feat: learning agent loop — self-improving agents via continuous learning pipeline#1327

Open

jeremyeder wants to merge 35 commits intoambient-code:mainfrom

jeremyeder:feat/learning-agent-loop

Commits on Apr 16, 2026

docs: add learning agent loop specs (002-009) with file-based architecture

Ambient Code Bot
and
claude
committed
docs: add continuous learning triage guide for knowledge curators

Ambient Code Bot
and
claude
committed
feat(backend,runner): add corrections realtime pipeline (spec 003)

Ambient Code Bot
and
claude
committed
feat: add "Correct This" UX for learning agent loop (spec 005)

Ambient Code Bot
and
claude
committed
feat: implement project memory store (spec 002)

Ambient Code Bot
and
claude
committed
feat(runner): session-scoped correction injection (spec 004)

Ambient Code Bot
and
claude
committed
feat(backend): implement post-session insight extraction (spec 009)

Ambient Code Bot
and
claude
committed
feat: add cross-session memory (spec 008)

Ambient Code Bot
and
claude
committed
feat(backend): add event-driven feedback loop (spec 007)

Ambient Code Bot
and
claude
committed
feat: implement visible attribution (spec 006) - memory citations, Learning tab, and dashboard endpoints

Ambient Code Bot
and
claude
committed
refactor: remove custom runner learned.py, use wiki compiler pipeline

Ambient Code Bot
and
claude
committed
fix(docs): correct frontmatter example and knowledge delivery description

Ambient Code Bot
and
claude
committed
fix(specs): update 006/008/009 to reference wiki compiler pipeline

Ambient Code Bot
and
claude
committed
Merge feat/lal-002 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-003 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-004 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-005 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-006 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-007 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-008 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-009 into feat/learning-agent-loop
Ambient Code Bot
committed
Merge feat/lal-triager-guide into feat/learning-agent-loop
Ambient Code Bot
committed
fix(docs): clarify progressive disclosure in triager guide

Ambient Code Bot
and
claude
committed
docs: add CL integration proposal for rfe-creator and assess-rfe

Ambient Code Bot
and
claude
committed
Merge feat/lal-triager-guide progressive disclosure fix
Ambient Code Bot
committed
fix: simplify pass — bugs, resource leaks, deduplication

Ambient Code Bot
and
claude
committed
fix(frontend): gate Project Memory and Learning nav items behind feature flag

Ambient Code Bot
and
claude
committed
fix(frontend): cast correction type keys for CORRECTION_TYPE_LABELS

Ambient Code Bot
and
claude
committed
fix(frontend): use enabled not data from useWorkspaceFlag

Ambient Code Bot
and
claude
committed
feat(frontend): add repo field to Add Memory dialog

Ambient Code Bot
and
claude
committed
feat(frontend): add repo history to Add Memory dialog

Ambient Code Bot
and
claude
committed
feat: add backend CreateLearnedPR endpoint for Add Memory

Ambient Code Bot
and
claude
committed

Commits on Apr 17, 2026

fix: improve GitHub auth error message for Add Memory

Ambient Code Bot
and
claude
committed
fix(frontend): improve Add Memory UX — no auto-open, validate URL, show PR details

Ambient Code Bot
and
claude
committed
feat(frontend): add repo config to Project Memory page

Ambient Code Bot
and
claude
committed