Remove template-only ABI files that falsely implied formal verification

Template Idris2 ABI files (Types.idr, Layout.idr, Foreign.idr) contained
only RSR template scaffolding with unresolved placeholders and no
domain-specific proofs. Removed to prevent false impression of formal
verification coverage.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

PROOF-NEEDS.md

@@ -0,0 +1,26 @@
+# Proof Requirements
+
+## Current state
+- ABI directory exists (template-level)
+- No dangerous patterns found
+- 100K lines; bot fleet for repository quality enforcement
+- Claims: "formal mathematical/statistical verification"
+
+## What needs proving
+- **Bot action safety**: Prove that automated bot actions (PRs, issues, fixes) are reversible and do not destroy repository state
+- **Confidence threshold correctness**: Prove that the robot-repo-automaton confidence thresholds correctly gate automated merges (no auto-merge below threshold)
+- **Rule conflict resolution**: Prove that when multiple bots (rhodibot, echidnabot, sustainabot, etc.) produce conflicting recommendations, the resolution is deterministic and documented
+- **Rate limiting**: Prove bots respect GitHub API rate limits and do not cause cascading failures across the fleet
+
+## Recommended prover
+- **Idris2** — For bot action safety and confidence threshold properties
+- **Lean4** — For algebraic properties of confidence scoring if modeled as an ordered semiring
+
+## Priority
+- **MEDIUM** — The fleet manages automated repository changes. Incorrect bot behavior at scale could damage many repos. The "formal verification" claim should be substantiated.
+
+## Template ABI Cleanup (2026-03-29)
+
+Template ABI removed -- was creating false impression of formal verification.
+The removed files (Types.idr, Layout.idr, Foreign.idr) contained only RSR template
+scaffolding with unresolved {{PROJECT}}/{{AUTHOR}} placeholders and no domain-specific proofs.

TEST-NEEDS.md

@@ -0,0 +1,92 @@
+# Test & Benchmark Requirements
+
+## Current State
+- Unit tests: ~22 Rust test files + shared-context tests — count unknown (no Cargo.toml at root to run)
+- Integration tests: partial (echidnabot/tests/integration_tests.rs, rhodibot/tests/integration_tests.rs)
+- E2E tests: NONE
+- Benchmarks: 5 files exist
+- panic-attack scan: NEVER RUN
+
+## What's Missing
+### Point-to-Point (P2P)
+205 Rust + 367 JavaScript + 42 ReScript + 34 Julia + 13 Haskell + 9 Idris2 + 85 Shell source files:
+
+#### Bots (major components):
+- **echidnabot/** — has integration tests, but coverage vs source count unclear
+- **rhodibot/** — has integration tests
+- **sustainabot/** — has test_sample.rs + .env.test, minimal coverage
+- **finishbot/** — has analyzers/testing.rs (test utilities, not tests OF finishbot)
+- **glambot/** — no tests found
+- **seambot/** — no tests found
+- **panicbot/** — no tests found
+
+#### Shared Context (Rust):
+- shared-context/tests/context_tests.rs — exists
+- shared-context/tests/fleet_coordination_test.rs — exists
+- Other shared modules may lack coverage
+
+#### Robot-repo-automaton:
+- No test files found
+
+#### Dashboard:
+- No test files found
+
+#### Campaigns:
+- No test files found
+
+#### Hooks:
+- No test files found
+
+#### Tasks/Scripts:
+- 85 shell scripts — no tests
+- 367 JavaScript files — no tests (these are likely bot action scripts)
+- 34 Julia files — no tests
+
+### End-to-End (E2E)
+- Bot fleet coordination: dispatch task -> bot processes -> report results -> aggregate
+- Individual bot lifecycle: start -> receive event -> analyze -> act -> report
+- echidnabot: receive webhook -> analyze repo -> generate findings -> submit
+- rhodibot: detect issue -> apply fix -> create PR
+- sustainabot: check dependencies -> evaluate health -> report
+- finishbot: analyze completion -> identify gaps -> notify
+- glambot: check style -> suggest fixes
+- seambot: check integration points -> verify
+- panicbot: run security scan -> report findings
+- Robot-repo-automaton: detect issue -> calculate confidence -> auto-fix
+- Fleet coordination: multiple bots on same repo -> deconflict
+
+### Aspect Tests
+- [ ] Security (webhook signature verification, bot credential handling, repo access scoping, auto-fix safety)
+- [ ] Performance (fleet throughput, bot startup latency, concurrent repo processing)
+- [ ] Concurrency (bot deconfliction, shared-context locking, parallel webhook processing)
+- [ ] Error handling (bot crash recovery, API rate limiting, malformed webhooks)
+- [ ] Accessibility (dashboard UI if applicable)
+
+### Build & Execution
+- [ ] cargo build per bot — not verified
+- [ ] Individual bot startup — not verified
+- [ ] Fleet orchestration — not verified
+- [ ] Dashboard build — not verified
+- [ ] Self-diagnostic — none
+
+### Benchmarks Needed
+- Per-bot analysis throughput (repos/hour)
+- Fleet coordination overhead
+- Webhook processing latency
+- Auto-fix confidence score accuracy
+- Memory usage per active bot
+
+### Self-Tests
+- [ ] panic-attack assail on own repo
+- [ ] Fleet health check
+- [ ] Per-bot self-test
+- [ ] Shared-context integrity verification
+
+## Priority
+- **HIGH** — Bot fleet (205 Rust + 367 JS + 42 ReScript + 34 Julia files) with tests only for echidnabot, rhodibot, and shared-context. 5 out of 7+ bots have ZERO tests. The 367 JavaScript files (bot action scripts) are completely untested. The robot-repo-automaton that auto-fixes repos based on confidence scores has no tests — this is a system that makes automated changes to other repos and needs extremely high correctness guarantees.
+
+## FAKE-FUZZ ALERT
+
+- `tests/fuzz/placeholder.txt` is a scorecard placeholder inherited from rsr-template-repo — it does NOT provide real fuzz testing
+- Replace with an actual fuzz harness (see rsr-template-repo/tests/fuzz/README.adoc) or remove the file
+- Priority: P2 — creates false impression of fuzz coverage