refactor: remove self-learning and the optimizer package (P1.4)

Self-learning is excluded from the community edition (ADR-0004); the
project owner reworks it separately. Cost tracking stays.

- agent: delete behaviour evolution (evolve-schema, iterative GEPA,
  compose-evolution, evolution gates, orchestrator and deploy),
  improvement and correction detection, AgentLearning, the eval-dataset
  builder, optimization-capture and the agent_insights tool. Secret
  detection moves from eval-dataset to secret-patterns. The LLM wiring
  keeps only the Haiku adapter and judge call used by skill audit and
  hygiene, so @ax-llm/ax is now declared by agent instead of server.
- server: delete /api/evolution/*, the evolution and optimizer services,
  GEPA expansion of the first chat message, the prompt_optimization and
  monthly_assessment jobs, the idle-session reviewer and its
  session-reviewer persona, prompt injection of past corrections and
  thumbs-down recorded as correction signals. Existing job rows of the
  removed types are ignored, not deleted.
- core: delete OptimizationLogStore; stop re-exporting the improvement
  signal and evolution run stores.
- web: delete the Evolution screen and the two job types in the
  automation editor.
- delete packages/optimizer.

Also logs UF-027 (self-learning loops read data nothing collects) and
UF-028 (the Settings prompt-shape choice is ignored by chat).
This commit is contained in:
shumeiko.it
2026-10-20 00:54:58 +02:00
parent b734c1d3d2
commit 2e1999a3f6
112 changed files with 336 additions and 16165 deletions

View File

@@ -1,19 +1,17 @@
/**
* HarnessTraceBridge — translates workflow-harness phase events into
* execution traces so harnessed agent runs feed the self-evolution loop.
* execution traces so harnessed agent runs show up in cost tracking and
* run history.
*
* Without this bridge the workflow harness is an island: phases run,
* gates pass or fail, but nothing flows into the `execution_traces`
* table that the evaluation dataset builder mines. This module closes
* that loop by subscribing to the shared `harnessEvents` emitter and
* writing one trace per phase outcome:
* table. This module subscribes to the shared `harnessEvents` emitter and
* writes one trace per phase outcome:
*
* harness:phase:complete → trace outcome 'verified'
* harness:phase:fail (aborted === true) → trace outcome 'abandoned'
* harness:phase:fail (will retry) → skipped — only final
* outcomes become training
* signal, mid-retry noise
* stays out of the dataset
* outcomes are recorded
*
* Each trace records:
* - input: the phase's instruction text