v0.12.0

Open

No due date

•Last updated Oct 5, 2026

Release v0.12.0: consolidation, tech-debt and documentation gaps. Entry: tech-debt, canonical, refactor and guards issues, documentation gaps, and bugs in infra, deploy, CI and test components. Never ahead of an open blocker.

10% complete

perf(rag): BM25 IDF uses query-local candidate set, not corpus stats (#11022 split)
mrveiss/AutoBot-AI#11064 · opened on Jul 6, 2026 · v0.12.0
1
canonical(chat): 'list chats' served by 3 overlapping endpoints — list_chats mounted at both /api/chats and /api/chat/chats, plus /api/chat/sessions as a 3rd envelope
mrveiss/AutoBot-AI#12371 · opened on Jul 24, 2026 · v0.12.0
canonical(slm-frontend): no backend contract at all — 0 client classes, 51/167 files raw-fetch, 23 hand-typed responses, NO type-generation against the 306-endpoint SLM backend (worse than #12363)
mrveiss/AutoBot-AI#12420 · opened on Jul 25, 2026 · v0.12.0
7
perf(knowledge): facts/by_category uses SMEMBERS+sorted() per request (no server-side SET pagination); legacy SCAN offset unstable
mrveiss/AutoBot-AI#12426 · opened on Jul 25, 2026 · v0.12.0
1
refactor(automation): consolidate WorkflowRunner + orchestration WorkflowExecutor capabilities onto canonical services engine (scoped follow-up to #12373)
mrveiss/AutoBot-AI#12577 · opened on Jul 26, 2026 · v0.12.0
tech-debt: relocate ProviderFallbackView/SystemHealthView from autobot-frontend into SLM
mrveiss/AutoBot-AI#12637 · opened on Jul 26, 2026 · v0.12.0
1
tech-debt: triage same-name class/enum clusters for consolidation (SourceType/SearchResult/ToolCall candidates) — round-5
mrveiss/AutoBot-AI#12771 · opened on Jul 27, 2026 · v0.12.0
10
tech-debt(backend): 51 of 95 health/status handlers await unbounded — #12751's guard test cannot land until they are triaged
mrveiss/AutoBot-AI#12892 · opened on Jul 28, 2026 · v0.12.0
tech-debt(analytics): endpoint scan still misses 274 real routes (93.1% of missing findings are false) after the registry-module fixes
mrveiss/AutoBot-AI#12953 · opened on Jul 29, 2026 · v0.12.0
5
perf(http): 134 raw aiohttp.ClientSession constructions remain (127 per-request) despite the shared pool at 151 sites
mrveiss/AutoBot-AI#12979 · opened on Jul 29, 2026 · v0.12.0
23
Planner does not tag sub-question topics as #12621/#12625 design assumed
mrveiss/AutoBot-AI#13015 · opened on Jul 30, 2026 · v0.12.0
Dead code: ai_hardware_accelerator._gpu_semantic_search and KnowledgeRetrievalAgent have zero production callers
mrveiss/AutoBot-AI#13028 · opened on Jul 30, 2026 · v0.12.0
1
bug(optimization): layer streaming loads the entire checkpoint per layer, per token (#13030)
mrveiss/AutoBot-AI#13031 · opened on Jul 30, 2026 · v0.12.0
4
bug(optimization): LayerInferenceEngine cannot produce correct output — no embedding, no LM head, wrong sampling axis (#13030)
mrveiss/AutoBot-AI#13032 · opened on Jul 30, 2026 · v0.12.0
3
bug(optimization): KV trim_to_length retains the oldest window instead of the newest (#13030)
mrveiss/AutoBot-AI#13033 · opened on Jul 30, 2026 · v0.12.0
5
tech-debt(frontend): SLM node-role vocabulary hand-mirrored, missing 'docker' role; OpenAPI pipeline can't carry it as-is
mrveiss/AutoBot-AI#13072 · opened on Jul 31, 2026 · v0.12.0
1
tech-debt(slm-frontend): 114 hand-declared types still shadow a generated OpenAPI schema (from #12420)
mrveiss/AutoBot-AI#13138 · opened on Jul 31, 2026 · v0.12.0
4
api/*.py consumers still bind MagicMock code_intelligence stubs at import time (precommit_analyzer, redis_optimizer, anti_pattern_detector)
mrveiss/AutoBot-AI#13181 · opened on Aug 1, 2026 · v0.12.0
1
docs(contributing): contributor path still steers PRs to main — default branch, duplicate PR template, unbased branch step
mrveiss/AutoBot-AI#13211 · opened on Aug 1, 2026 · v0.12.0
1
MCPBridgeManifest.resource_limits is declared and plumbed but set by no bridge
mrveiss/AutoBot-AI#13229 · opened on Aug 1, 2026 · v0.12.0
1
test(knowledge/backends): ChromaDB contract coverage is stubbed out — decide how to restore it (follow-up to #13239)
mrveiss/AutoBot-AI#13242 · opened on Aug 1, 2026 · v0.12.0
umbrella(config): #7437 migration silently dropped ~48 production defaults — fourth one-off fix, no systemic tracking
mrveiss/AutoBot-AI#13264 · opened on Aug 1, 2026 · v0.12.0
5
ci: check-ts-delta enforces a stale 188-error baseline behind a step that already enforces zero
mrveiss/AutoBot-AI#13353 · opened on Aug 3, 2026 · v0.12.0
llm_shared/optimization: two implementations of meta-device layer eviction, one hand-rolled
mrveiss/AutoBot-AI#13404 · opened on Aug 3, 2026 · v0.12.0
1
tech-debt(code-intelligence): code_indexer's tree-sitter passes don't parse imports, so cross-module resolution never gets import-context confidence
mrveiss/AutoBot-AI#13491 · opened on Aug 4, 2026 · v0.12.0

Navigation Menu

v0.12.0

perf(rag): BM25 IDF uses query-local candidate set, not corpus stats (#11022 split)

canonical(chat): 'list chats' served by 3 overlapping endpoints — list_chats mounted at both /api/chats and /api/chat/chats, plus /api/chat/sessions as a 3rd envelope

canonical(slm-frontend): no backend contract at all — 0 client classes, 51/167 files raw-fetch, 23 hand-typed responses, NO type-generation against the 306-endpoint SLM backend (worse than #12363)

perf(knowledge): facts/by_category uses SMEMBERS+sorted() per request (no server-side SET pagination); legacy SCAN offset unstable

refactor(automation): consolidate WorkflowRunner + orchestration WorkflowExecutor capabilities onto canonical services engine (scoped follow-up to #12373)

tech-debt: relocate ProviderFallbackView/SystemHealthView from autobot-frontend into SLM

tech-debt: triage same-name class/enum clusters for consolidation (SourceType/SearchResult/ToolCall candidates) — round-5

tech-debt(backend): 51 of 95 health/status handlers await unbounded — #12751's guard test cannot land until they are triaged

tech-debt(analytics): endpoint scan still misses 274 real routes (93.1% of missing findings are false) after the registry-module fixes

perf(http): 134 raw aiohttp.ClientSession constructions remain (127 per-request) despite the shared pool at 151 sites

Planner does not tag sub-question topics as #12621/#12625 design assumed

Dead code: ai_hardware_accelerator._gpu_semantic_search and KnowledgeRetrievalAgent have zero production callers

bug(optimization): layer streaming loads the entire checkpoint per layer, per token (#13030)

bug(optimization): LayerInferenceEngine cannot produce correct output — no embedding, no LM head, wrong sampling axis (#13030)

bug(optimization): KV trim_to_length retains the oldest window instead of the newest (#13030)

tech-debt(frontend): SLM node-role vocabulary hand-mirrored, missing 'docker' role; OpenAPI pipeline can't carry it as-is

tech-debt(slm-frontend): 114 hand-declared types still shadow a generated OpenAPI schema (from #12420)

api/*.py consumers still bind MagicMock code_intelligence stubs at import time (precommit_analyzer, redis_optimizer, anti_pattern_detector)

docs(contributing): contributor path still steers PRs to `main` — default branch, duplicate PR template, unbased branch step

MCPBridgeManifest.resource_limits is declared and plumbed but set by no bridge

test(knowledge/backends): ChromaDB contract coverage is stubbed out — decide how to restore it (follow-up to #13239)

umbrella(config): #7437 migration silently dropped ~48 production defaults — fourth one-off fix, no systemic tracking

ci: check-ts-delta enforces a stale 188-error baseline behind a step that already enforces zero

llm_shared/optimization: two implementations of meta-device layer eviction, one hand-rolled

tech-debt(code-intelligence): code_indexer's tree-sitter passes don't parse imports, so cross-module resolution never gets import-context confidence

Issues