MOTO v1.0.8#38
Merged
Merged
Conversation
added 30 commits
April 8, 2026 17:36
…acking, and UI updates
… improvements, bug fixes
# Conflicts: # frontend/src/components/BoostLogs.css # frontend/src/components/BoostLogs.jsx
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Features - Lean 4 Integrated: Automated theorem generation pipeline completed with Lean 4 proof verification as the final validation step, this new mechanic includes the MOTO ASI's primary Top-P exploration mechanic for feedback/retry and deeper exploration on proof retry attempts. - New autonomous proof pipeline modules: proof identification, lemma search, proof formalization, dependency extraction, novelty checking, proof database, and a dedicated proof verification stage. - New proof surfaces in the GUI: Mathematical Proofs panel, Proof Graph visualization, and a Proof notification stack. - New backend routes: `/api/health`, `/api/features`, `/api/update-notice`, and a full proofs API (proof database, Lean 4 status, manual proof checks, certificate export). - Docker support: first-class `Dockerfile`, `.dockerignore`, and `docker/entrypoint.sh` for hosted/generic-mode deployments. - Ubuntu 24.04 launcher: new `Launch MOTO.sh` bootstrapper that provisions a repo-local `.venv` and runs the shared Python launcher. - Launcher and updater infrastructure rewritten in Python (`moto_launcher.py`, `moto_updater.py`) with a new `moto-update-manifest.json` describing version, build commit, and API contract version for update comparison. - Generic/hosted mode added, with `requirements-generic.txt`, hosted-proxy authentication helpers (`backend/api/proxy_auth.py`), a Lean-free / Z3-free hosted contract, and a dedicated `fastembed` embedding provider for hosted environments. - Auto-fill OpenRouter context window from model context length. - Auto-cap OpenRouter max output to the lower of 20% model context or the smallest available host output cap. - The program now supports parallel program runtimes. - New automated test suite under `tests/` covering the Lean 4 client, launcher/updater, rigor Lean placement validator, compiler marker visibility, Wolfram tool loop, paper memory appendix, and more. ## Changes - Kimi K2.6 replaced Kimi K2.5 in any respective default profiles. - DeepSeek V4 Pro replaced DeepSeek V3.2 in the respective default profiles using DeepSeek. - Inception 2 Mercury added as a supplementary 4th parallel brainstorm submitter in the fast affordable profile. - Applied the new OpenRouter auto-fill and host-cap logic across Aggregator, Compiler, Autonomous Research, and Boost settings. - Added concise rule coverage for proof verification checkpoints, manual proof checks, proof APIs/events, session-aware proof storage, and proof graph/dependency surfaces. - Documented Lean 4 / SMT runtime gating, Lean-authoritative verification, and the hosted Lean-free/Z3-free contract. - Updated `.gitignore` with `backend/data/proofs/` (keep-pattern), `backend/data/lean4_workspace/` (blanket), and `backend/data/elan/` (blanket) to keep the Lean toolchain installer and workspace out of the repo, plus a new `backend/data/proofs/.gitkeep`. - Balanced fast submitters with slow submitters; any individual submitter role that has more than 4 submissions in the queue pauses while the others continue. - Added 2 "lab grade" tier default profiles featuring SOTA models such as Opus 4.7, GPT 5-5 Pro, and Grok 4.20 Multi-Agent. Runtime cost estimates range from $20 per hour to $60 per hour or greater for each respective profile. - Replaced the legacy `_moto_internal_launcher.ps1` bootstrap with the cross-platform `moto_launcher.py` entrypoint. - Trimmed the MIT LICENSE appendix (the non-legal AI-generated-content advisory block that followed the `---` divider); the MIT license text itself is unchanged. ## Bug Fixes - OpenRouter provider lookups now return endpoint metadata instead of only host names. - Fixed LM Studio "offline" status bug in GUI which showed LM Studio offline when it was not. - Silenced the ECONNREFUSED noise during startup while preserving real errors. - Fixed disclaimer GUI resolution bug discovered by @asimog. - Added GUI UIX optimizations proposed by @asimog. - Fixed tab switching screen orientation bug.
## Bug Fixes Fixed Matlab /git bug caused by excessive timeout.
## Bug Fixes Fixed Matlab /git bug caused by excessive timeout.
added 3 commits
May 5, 2026 09:33
## Bug Fixes Fixed Matlab and Lean 4 co-dependence bug that was disrupting some users proof generation.
## Bug Fixes Fixed Matlib and Lean 4 co-dependence bug that was disrupting some users proof generation.
## Features - Added LeanOJ Proof Solver as a third standalone mode with dedicated model profiles, cumulative brainstorming, proof storming, skip-brainstorm control, and Lean 4-verified final template solving. - Added hosted/generic-mode runtime support with an API-only sandbox contract, FastEmbed embeddings, OpenRouter-only inference, and capability flags for the hosted web wrapper. - Added pruned-paper preservation and history downloads so context-excluded Stage 2 papers remain reviewable by users. - Added "supercharge" mode that incorporates a mini-version of Top-P exploration at the individual model level. Increasing a model's intelligence before its final constituent answer within the harness itself. This has the largest impact on LM studio and budget model users that are running smaller models. - The brainstorm mechanic for all modes can now produce Lean 4 proofs in the brainstorm stage if relevant to the users prompt. ## Changes - Optimized proof-solving prompt flow to work directly toward the user prompt/template instead of only building exploratory context. - Added statement-alignment validation so Lean-accepted proof code must correspond to the intended theorem candidate before storage or placement. - Added shared proof registration and integrity gates across autonomous proof verification, compiler rigor mode, and LeanOJ. - Updated project rules to document hosted runtime behavior, LeanOJ, and shared proof-integrity architecture. - Added shared post-Lean proof integrity checks for autonomous paper proofs, compiler rigor mode, and LeanOJ. - Compatible models screen is now on all settings. - Added "Major mathematical discovery" validator ranking category for proofs and relabeled the gold category to "Minor Mathematical Discovery". Users are now able to tell which discoveries are major vs minor as per their validator model's direction injection review. - Optimized and audited .cursor rules folder to assist with vibe coders and forks. - Parallel ran harness roles now utilize a temperature ladder such as 0, .2, .4, and .8 for concurrent LM studio API calls on models loaded with multiple slots. - Expanded OpenRouter model/provider settings, raw settings tooling, and automated context settings to avoid provider outliers that disrupted automatic context sizing. - Improved API logging, boost routing visibility, and live activity UI for long-running workflow state. - Improved autonomous proof verification and paper context handling during brainstorm-to-paper and resume flows. - Improved compiler rigor, critique, and Wolfram tool flows. - Updated launchers, updater metadata, and Linux hosted entrypoint behavior. - Refreshed documentation, rules, ignore files, and test coverage. ## Bug Fixes - Hardened desktop API access with loopback defaults, token-authenticated HTTP routes, one-time WebSocket tickets, and CSRF-style origin checks. - Hardened hosted proxy auth with signed body-hash headers so unauthenticated request bodies are not buffered before rejection. - Hardened workflow startup, provider error handling, and log redaction paths for safer default and hosted operation. - Fixed Boost secret persistence so OpenRouter keys are no longer written to boost state and restored Boost reports when a key is needed. - Locked Z3 executable configuration to trusted startup settings and rejected runtime `z3_path` proof-settings input. - Hardened final-answer archive access against nested paper/brainstorm path traversal. - Hardened PDF export by sanitizing server-side HTML, enforcing size caps, disabling JavaScript, and blocking external Chromium requests. - Redacted API/Boost logs by default so full prompts, responses, Wolfram tool results, and legacy full payloads are not exposed through log routes. - Sanitized OpenRouter/provider error bodies before logging or surfacing them to avoid leaking echoed prompts or keys. - Removed OpenRouter API keys from model-list query strings and require headers or stored runtime keys instead. - Disabled hosted self-update mutation and made workflow start checks atomic across top-level modes. - Reduced proof-status path disclosure by returning safe Lean/Z3 path labels instead of absolute local paths. - Fixed retry transcript hygiene so private model thought/channel tokens cannot poison retries, memory, or RAG context. - Fixed Mathlib and Lean 4 co-dependence bug that disrupted some users' proof generation. - Rejected fake proof devices (`axiom`, `constant`, `opaque`) introduced by generated Lean code unless already present in the source/template context. - Fixed LeanOJ skip-brainstorm handling so the skip request is not consumed before the brainstorm phase. - Fixed generic API helper path handling so configured API bases are respected. - Fixed API call logs so standard model calls are recorded alongside boosted calls. - Fixed OS hover pop-up layering showing behind neighboring boxes. - Fixed OpenRouter credit exhaustion causing proofs to be skipped if the user ran out during the proof creation step. - Fixed RAG lock release on cancelled embedding/indexing tasks so resumed research cannot stall. - Fixed autonomous paper writer resume so completed brainstorms restart at the proof/paper handoff instead of zero. - Fixed manual Lean 4 proof checks so they use the currently selected role models instead of stale brainstorm or paper generation models. - Fixed Lean 4 proof completion UI so finished proof passes show accurate live activity counts. Authored by Patrick White, Patrick@Intrafere.com
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
added 2 commits
May 17, 2026 22:38
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
MOTO v1.0.8
Features
Changes
Bug Fixes
z3_pathproof-settings input.axiom,constant,opaque) introduced by generated Lean code unless already present in the source/template context.Authored by Patrick White, Patrick@Intrafere.com