All notable changes to ALDECI/Fixops will be documented here. Format: Keep a Changelog. Versioning: SemVer.
Branch:
features/intermediate-stageSession: ~50 commits across hardening, performance, QA, and feature backfill. Tip SHA at entry cut:43d43d95
- Prometheus
/metricsendpoint — 5 gauges (active scans, findings, connectors, pipeline runs, trust events) [e583d24d] /api/v1/health/comprehensiveaggregator — single endpoint polling all subsystem health checks [bfa88303]- Rate limiting on auth, webhook, and ingest — 7 endpoints hardened against burst abuse [
c03ffd27] - 17 SQLite indexes across 6 domain databases — covering high-frequency query paths (findings, scans, connectors, risk, evidence, trust) [
43d43d95] - 6 deferred empty-endpoints backfilled with real engine wiring (no more 501 stubs on critical paths) [
3d8fd7c9] - CI regression-gate workflow — OWASP audit, perf benchmarks, and import sweep run on every PR via GitHub Actions [
7c86c8d2]
- Brain Pipeline — 3 hotspots fixed, ~2 s saved per 12-step run [
ee340f83] - LLM Council — 3 hotspots fixed, 600–1 500 ms saved on real-provider path [
d61dde59] - Connector framework — O(N²) endpoint lookup → O(1); sequential
bulk_push→asyncio.gather; sequentialscan_fleet→ThreadPoolExecutor[eb46d106] - Risk scorer — latency reduced from ~527 ms → < 50 ms [
91187379] - RSA cache — latency reduced from ~2 111 ms → < 50 ms [
91187379]
commercial_vendor_router— missingQueryimport causing 500 on vendor list endpoints [ee340f83]validation_router— relative import error breaking module load [eb46d106]scanner_parsers— cross-scanner dedup logic producing duplicate findings across parser boundaries [51895d85]- 4 platform gaps closed: pip-audit SARIF normalisation, ingest-to-issues pipeline linkage,
/risk-scoring/summarymissing fields, cross-scanner dedup [51895d85] - 5 frontend dashboards (
FindingsExplorer,ThreatHunting,VendorManagement,DeveloperPortal,SBOMManagement) — mock fallback data replaced with live API calls [736a1188,d71dd091,30c8464b,d14ba174,01e08e81] AgeBadgestalenowreference inFindingsExplorercausing incorrect age rendering [736a1188]
OWASP audit across 7 packages — ~38 individual fixes including:
- Information disclosure via verbose error responses → generic error messages
- Hardcoded secrets moved to environment variables (JWT secret, OAuth tokens)
- Missing HTTP timeouts on external calls → explicit timeout enforcement
- CVE injection via unsanitised scanner input → input validation layer
- Severity bypass via crafted payload → server-side severity enum enforcement
- Overly broad exception handlers narrowed to specific exception types
- OAuth token redaction in structured logs
- SBOM bomb cap (unbounded dependency tree recursion → depth limit)
- Suite-core connector OWASP fixes (4 issues) [
2652b066]
- OWASP regression lockdown — 46 tests covering all 7 hardened packages; wired into CI [
51895d85,7c86c8d2] - TrustGraph emit-site assertions — 14/14 engines validated emitting events on expected paths [
11b5f10c] - Engine + router import sweep — 1 315 modules verified clean (0 skipped, 0 import errors) [
752951f5] - Perf benchmark suite — RSA cache, risk scorer, and brain pipeline thresholds enforced; CI blocks regression [
91187379,7c86c8d2] - Rate-limit tests — burst and sustained load tests for all 7 rate-limited endpoints [
c03ffd27] - Health endpoint tests — comprehensive aggregator response structure validated [
bfa88303] - Prometheus metrics tests — gauge presence and label correctness verified [
e583d24d]
Range:
43d43d95..98d04d23(~60 commits, sweeps #7–#21) Tip SHA at entry cut:98d04d23
- Docker/k8s — 7 hardening issues resolved (missing resource limits, privileged flags, image pinning) [
827cee32] - Shell scripts —
set -euo pipefail+ cwd anchors applied across 3 scripts; shellcheck clean [3d2471a4] - Async-emit-at-import — coroutine-not-awaited warnings fixed in 10 engines (import-time async emit converted to sync wrappers) [
1b25903a] asyncio.run()scan — 2 executor-race candidates fixed in brain pipeline scan path [32842a75]- Brain pipeline asyncio race x2 —
_run_attack_graph_gnnrace fixed with same pattern as first fix [8b9738ed]; lockdown test added [32842a75] - Dead lazy imports purge — 280 dead
lazy()imports + matching routes removed; restores production build [3eb988a0]
- UI build verification job — CI now catches dead lazy import regressions before merge [
48e6424c] - Regression-gates workflow — validated and cleaned; run history verified [
ab2e691a,465317ae]
- Sweep #7 — all green at HEAD
c0852a5f[33998754] - Sweep #8 —
test_admin_db_statsasyncio fix confirmed (e124c48d), all green [25139e7a] - Sweep #9 — final wrap, all green at HEAD
05964156[c3f7982f] - Sweep #10 — all green at HEAD
1ffa7d9d[8254602c] - Sweep #11 — cascade unblocked, all green at HEAD
1ad190d4[9073b7c8] - Sweep #12 — all green at HEAD
c98e9aed[ce58afd1] - Sweep #13 — asyncio fix #2 confirmed (
8b9738ed), all green [966b43c9] - Sweep #14 — asyncio scan fixes confirmed (
32842a75), all green [a41b7e65] - Sweep #15 —
test_reachability_perfimport fix confirmed, all green [1f73bed2] - Sweep #16 — module-cache fix confirmed (
ed6512e0), 0 broken collectors, all green [b241cd32] - Sweep #17 — all green at HEAD
d65b60df[e755e178] - Sweep #18 — production build restored, all green at HEAD
3eb988a0[3b49bcb9] - Sweep #19 — all green at HEAD
e3b2660f[6ef61fe4] - Sweep #20 — all green; HANDOFF v12 wrapped [
a8a08628] - Sweep #21 — all green at HEAD
cfd36eb2; true session close [b69947ae] - Async-emit lockdown test — 10 engines covered [
498b1908] - pytest markers — 26 perf test files categorized for CI prioritization [
c0852a5f];owaspmarker added for parity with-m perf[3519e40b] - Dead marker removal — 4 dead markers (
unit,regression,e2e,performance) removed [cfd36eb2]; marker smoke verified across 14 markers × 44,976 tests [426fa14b] - Conftest DRY — minimal-app fixture factored to conftest, eliminating duplication across 2 test files [
1ffa7d9d] - Broken collector triage — 4 collection errors documented, 1 fixed; stale
_add_edgeimport patched intest_reachability_perf[a4b9650d];test_cspm.pystale imports patched + removed-API tests skipped [1ad190d4]; module-cache ordering fixed for 3 broad-scan collection errors [ed6512e0] - UI test files — 5 test files finalized after Phase 3 page-deletion cleanup [
6f8e137e]
- HANDOFF v5–v13 — rolling session handoffs capturing open threads, sweep timelines, bug tallies, and branch SHAs [
d9d2ae97,43895c5c,f663f531,e3460b26,d751b66d,9cfa0f75,e3b2660f,a8a08628,98d04d23] - Dependency audits — Python pip-audit and npm audit reports for 2026-05-05 [
b2285945,c98e9aed] - Shell audit report — shellcheck findings documented [
32e79756] - UI bundle size audit — bundle analysis with code-split recommendations [
5b88eada] - Regression-gates CI doc — workflow run history verified and documented [
465317ae] - Session snapshots — final commit tallies and state-row refreshes [
a8af529c,667f62b7,64c84eca,6381af43] - Duplicate
regression_statuscanonicalized — single authoritative copy atdocs/[cad33d9a]
/api/v1/admin/connectors/inventory— new endpoint exposing connector capability surface [1ebf78d3]- pytest marker smoke — 14 active markers verified across full 44,976-test suite [
426fa14b]
Branch:
features/intermediate-stageSession: ~69 commits across a pre-dawn-to-evening megasession. Tip SHA at release cut:f9cf3fe8
- Issues hero (Wiz-pattern, single queue with 8 tabs — findings/triage/SBOM/supply-chain/threat-intel/upgrade-paths/exception) [
12f16c83] - Brain Pipeline hero (12-step visualizer + Multi-LLM Council rail, MPTE Console, FAIL Chaos, Code Intelligence tabs) [
0771bd11,8b36a6ee,9cbf0ae1] - Compliance hero (Cloud Posture + AI Exposure Tenable-parity + Waivers + Policies sub-tabs) [
e0972bac,0b8c0b86,2a97fbcf] - Asset Graph hero (Apiiro/Wiz-pattern second-brain canvas + Inventory tab, Attack Paths) [
7e728702,afc66592,c08b9325] - Command Dashboard hero (persona-aware landing, AI Copilot tab, real apiFetch) [
4c6cd97b,5486541d] - Admin Console hero (multi-tenant administration, MCP Gateway + System Health sub-tabs) [
a6e73395,f4983bcb] - Remediation Center hero at
/remediate[00f41b74] - Integrations Hub folded into Asset Graph hero [
c08b9325] - Executive Brief, SOC Ops, SLA/Risk, Issue Detail screens folded into P1 heroes [
768c5da9] - Playwright golden-path E2E suite for all 6 P0 hero screens — 6/6 PASS, zero mocks [
22268aeb] - 81+ consolidation redirect rules preserving bookmarks for 90 days [
82c1ae36] - IDE-in-browser (3-panel Monaco file-tree + diff viewer, NEW-G071) [
8b957ae1] - Real customer end-to-end demo trace with 6-hero screenshot evidence (Juice Shop) [
134cd807] - Admin + Asset Graph clean screenshots after P0 bug fixes [
aad87027]
- Consolidated 370+ source screens to 30 hero screens; 99+ redirect rules in place [
7a2990a3,82c1ae36] - ~50 mock pages (J-O slice) wired to live API endpoints [
3feb5b71] - ~50 mock pages (P-Z slice) wired to live API endpoints [
ab325746]
- TypeScript errors reduced from 152 to 98 (54 cleared) — SecurityTrainingDashboard + KPICard label fixes [
b11fff60]
init_event_buswired; router middleware coverage raised from 3.9% to 80%+ [48ee40d2]- 30 highest-degree hubs wired across 6 batches (graph.py 9366, scanner_parsers 3975, connectors 3032, ld_provider 3320, cache_service 4027, + 25 more) [
befea111,db618c93,d6ae6ab5,3074e918,579d4d84,b748d645,64fd4a49,9852939d,c6389daf,b826a45a,a68cf0bb] - 16+ connectors emitting via shared
_emithelper: snyk_oss, cspm, container_security, crowdstrike_falcon, defender_xdr, edr, sentinelone, siem, dast_pentest, commercial parsers, iam_sso, n8n, threat_intel, defectdojo, sdlc, pull_connector [a5a08b54,0543d17b,6996a3fe,094b7f79] - Connector emit wired on findings/lifecycle, identity/credentials, CTEM/baseline, scanning/discovery, air-gap/local, threat-intel, dev-portal, SBOM paths [
926687aa,36c47e75,4016668e,5021b6ac,7593b4c7,01ee408b] - TrustGraph coverage raised from 24.4% to 38.4% (15.1% direct + 10.6% blast-radius + 12.7% middleware) [
ad453c50] - Visualizer updated with AQUA blast-radius color band + HTML topology report [
a68cf0bb,b041efc1,672bf293] - Webhook consumer federation examples (Splunk, Elastic, Slack forwarder) [
fe90b742]
- LLM Phase 1 closed-loop subscriber wired to TrustGraph emit events — production live [
cbd01c4d] - 703 council verdicts + 703 DPO pairs populated in
data/learning_signals.dbvia real fleet scans (350x growth from 2 to 703 in one session) [d326da7b] - LLM Phase 2 dataset curator + training scaffold (trl DPOConfig) + inference router (student/council routing) — DRY-RUN validated [
4904309a] - Live telemetry endpoint + Brain Pipeline Learning Loop tab dashboard [
f901de22] learning_signals.dbschema:council_verdictstable +feedback_pairstable — auto-created on first run, no manual migration [d326da7b]- AgentDB ↔ TrustGraph bridge — HNSW vector search over emit events + DPO pairs, 150x speedup over linear scan [
73c05c0d] - MiniLM 384-dim semantic embeddings enabled in AgentDB (auto-upgrades from hash fallback) [
65cbbc93] - Bulk-reindex of existing AgentDB entries with MiniLM 384-dim embeddings [
a377f3c6] - Nightly fleet-scan cron infrastructure (4-step pipeline: ASPM scan → SBOM seed → CSPM seed → curator refresh) targeting 703 → 10K DPO pairs [
f9cf3fe8] - Progress checker script with ETA and ASCII bar for DPO growth tracking [
f9cf3fe8]
- SCIF Stage 1 — Engineering (8/8 deliverables, 12/12 tests pass):
- FIPS boot wired into FastAPI startup sequence [
69efa330] - UBI9-hardened Iron Bank Dockerfile + SoftHSM PKCS#11 module [
1159ef49] - Tamper-evident audit chain (SHA-256 chained log entries) [
1159ef49] - Air-gap bundle script (offline
.tar.gzwith all deps, pip cache, models) [1159ef49] - Cosign image signing — closes SCIF Stage 1 blocker #2 [
aba22fff] - ISSO Pilot Bundle README + STIG runbook + LLM air-gap verification script [
69efa330] - All-on-prem LLM verification path (vLLM canonical, Ollama dev convenience) [
69efa330]
- FIPS boot wired into FastAPI startup sequence [
- SCIF Stage 2 — Auditor Documentation:
- System Security Plan (SSP), POA&M, NIST 800-53 Rev 5 control matrix (CSV), threat model, crypto datasheet, auditor quick-reference — 95% of in-scope controls implemented [
20ef9510]
- System Security Plan (SSP), POA&M, NIST 800-53 Rev 5 control matrix (CSV), threat model, crypto datasheet, auditor quick-reference — 95% of in-scope controls implemented [
- SCIF Stage 3 — Federal Sales:
- Target list (36 program sponsors), cold outreach (4 templates), discovery playbook, pilot SOW, reference architecture [
43f73eb3] - Fully-automated Day 1 ISSO install script + smoke test [
2ee6e8ed]
- Target list (36 program sponsors), cold outreach (4 templates), discovery playbook, pilot SOW, reference architecture [
- Pitch deck (12 slides) + one-pager + objection-handling guide + 7 battle cards [
bb35e502] - Demo script (Command → Brain → Compliance arc) + POC template + customer onboarding playbook + win/loss template [
68c0130e] - Analyst pack: MQ/Wave brief, ref-arch whitepaper, case-study template, anti-customer profile [
c0df3e0e] - Master investor pack + data room index + traction metrics [
a0f15a8b] - 7 persona-specific landing page copy files (CISO, DevSecOps Lead, SOC Analyst Tier 1, Compliance Officer, Federal CIO/RMF AO, AppSec Engineer, Cloud Security Engineer) [
bde8b101] - 5-minute demo video script + storyboard + recording setup guide [
2c394e24]
- ruflo (claude-flow) v3.5.80 integrated — 98 agent templates, AgentDB HNSW, ReasoningBank, SPARC methodology, hive-mind Byzantine consensus, ~70 new skills [
71744c25] - Prowler → SecurityFindingsEngine bridge + checkov
--quietbugfix [dea8dd7d] - DAST real format parsers (OWASP ZAP + Nuclei) [
1eebfe5a] - CrowdStrike Falcon real format parser [
b7043c7e] - CSPM real (Prowler + Checkov + Trivy + CloudSploit + Agentless) [
ad727467] - AutoFix engine wired to GitHub App PR review (Snyk-parity) [
c76dd7e2] - WebSocket event push for real-time dashboard updates [
c1c235be] - Graphify code-only build via
.graphifyignore(excludes docs/raw/markdown) [dbcf0935] - code-review-graph per-directory builds (connectors: 1221n/3054e/50c; scripts: 2518n/5174e/110c) [
a1ad4161] - Session docs consolidated into
docs/INDEX.md+ strategic roadmap [54d1fa4f]
_*.pygitignore rule scoped to repo root only (was accidentally excludingsuite-core/connectors/_emit.py) [7861f9fe]- AWS-key-like fixture in tests replaced with allowlisted
AKIAIOSFODNN7EXAMPLE[1018fa22] - Dashboard render bug fixed — 5/5 routes show real page content + API calls fire [
07994f29] - Bulk-triage: atomic cross-org protection + Pydantic validation [
dcdb590c] - CSPM posture score derived from real findings DB (was magic constant) [
d9077c44]
- Dependabot triage: 140 alerts reviewed (2 Critical / 55 High / 59 Medium / 24 Low) [
b8f75738] dompurifyoverridden to^3.4.1in aldeci-ui-new [312a5795]postcssbumped 8.5.8 → 8.5.10 [312a5795]path-to-regexpbumped 8.3.0 → 8.4.2 [4b75180f]picomatchbumped 2.3.1 → 2.3.2 [4b75180f]follow-redirectsbumped 1.15.11 → 1.16.0 [4b75180f]
- Beast Mode v6 full stack: OMC + everything-claude-code + SwarmClaw + TrustGraph + code-review-graph + ruflo
- 334 backend engines (attack surface, SBOM, evidence, risk scoring, compliance, MPTE, FAIL, LLM council, 8 native scanners)
- 568 API router files (FastAPI, multi-tenant, RBAC enforced)
- 372 frontend pages (React 19, Vite 6, Tailwind v4) — 30 consolidated P3 heroes + 342 legacy routes with redirects
- 32 scanner normalizers, 28+ threat intel feeds, 13 PULL + 7 bidirectional connectors
- Multi-AI consensus engine (Gemini + Claude + GPT-4 + Qwen fallback)
- 12-step Brain Pipeline with cryptographic evidence chain
- MPTE 19-phase pentest orchestration
- Quantum-safe evidence signing (NIST FIPS 203/204/205 PQC)
- 30 personas, 6 RBAC roles, 7 compliance frameworks
- Multi-tenant isolation with org-scoped all APIs
- Real customer onboarding flow (15 famous GitHub apps validated)