coulomb/agentic-resources
2
0
Fork 0
generated from coulomb/repo-seed
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
39 Commits 1 Branch 0 Tags
7646cbc358181216593cd2ecc17198eac035e04b
Commit Graph

9 Commits

Author SHA1 Message Date
tegwick
1b6081cd88 session-memory: denoise error fingerprints (WP-0006 follow-up) 
Tighten _is_failed: exclude successful hub JSON responses (top-level no-error
payloads) and file-read snapshots (numbered cat -n source lines) that were
polluting error_snippets. JSON verdict classifies error vs success payloads
directly. Cuts distinct fingerprints 444 -> 269 (~40%) over the real corpus with
the top errors unchanged. Assessment caveat updated. 5 new tests; suite 102/102.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-07 13:39:08 +02:00
tegwick
97379e9658 session-memory: error-body mining into digest (WP-0006 T01) 
build_digest now extracts normalized error fingerprints + samples from failed
events (error kind + failing tool_result bodies) into a durable error_snippets
list — paths/numbers/uuids/addrs stripped so the same error collapses to one
fingerprint with a count; Python traceback header skipped in favour of the real
exception line. Durable in Tier 2 (survives Tier 1 eviction). SCHEMA_VERSION ->
2 (re-ingest needed to populate). 7 new tests; suite 95/95 green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-07 12:45:01 +02:00
tegwick
436a96dcd8 session-memory Phase 1: Detect pipeline (T04-T07) 
- detect/signals.py: pure extractors over digests (retry storm, repeated
  errors, budget overrun vs corpus p90, abandoned, clean pass, recovery)
- detect/cluster.py: deterministic clustering into candidate Patterns with
  evidence (sessions/repos/flavors/cost impact) + cross-flavor flagging
- detect/__main__.py: python -m session_memory.detect, ranked report
  (cross-flavor first) + --json; persists candidates to Tier 2 patterns table
- core/store.py: list_digests + save_patterns
- tests for signals, cluster, detect entrypoint

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 22:31:13 +02:00
tegwick
bc11cb9aec session-memory Phase 1: Codex adapter (T01) + multi-file merge (T03) 
- adapters/common.py: shared Normalized + helpers (resolve_repo, classify_tool,
  jsonl iter, etc.); claude.py refactored to use it (Normalized re-exported)
- adapters/codex.py: rollout {timestamp,type,payload} parser; session_meta/
  response_item/event_msg mapping; flat call_id join; token_count cost;
  registered in ingest dispatch
- core/store.py: ingest() now merges multi-file sessions by content
  fingerprint, appends new events with offset seq (design OQ6); idempotent
- tests/test_codex_adapter.py, tests/test_merge.py

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 21:55:32 +02:00
tegwick
586ed90948 session-memory Phase 0: ingest cursor + sweep entrypoint + config (T06) 
- session_memory/core/cursor.py: size/mtime change detection sidecar
- session_memory/config.toml: store paths, retention caps, per-source
  globs (claude on, codex/grok off for Phase 1), repo->domain map
- session_memory/ingest.py: discover->normalize->store->digest->evict;
  --dry-run creates/writes nothing; python -m session_memory.ingest
- tests/test_ingest.py; live dry-run parsed 84/85 real local sessions

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 21:41:59 +02:00
tegwick
451fb8f1f3 session-memory Phase 0: budget-based retention sweep (T05) 
- session_memory/core/retention.py: RetentionConfig + sweep() with backstop,
  budget (oldest-analyzed-first, never touches un-analyzed), and hard-cap
  overflow (analyze-now then reported last-resort data_loss); EvictionReport
- tests/test_retention.py covers all four branches

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 21:37:40 +02:00
tegwick
abb888f3ef session-memory Phase 0: session digest + outcome heuristic (T04) 
- session_memory/core/digest.py: build_digest (cost totals, kind/tool
  histograms, markers, snippets) + cross-flavor infer_outcome heuristic;
  analyze() promotes Tier1->Tier2 and sets analyzed_at (-> evictable)
- tests/test_digest.py

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 20:03:04 +02:00
tegwick
29fc211a14 session-memory Phase 0: Tier1/Tier2 store (T03) 
- session_memory/core/store.py: SQLite rows + blob-dir bodies, idempotent
  ingest on (session_uid,seq), Tier1/Tier2 usage accounting, evict_raw that
  drops raw but preserves the digest; watermark columns authoritative
- tests/test_store.py: ingest idempotency, accounting, eviction invariant

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 19:10:02 +02:00
tegwick
1c29a94fa9 session-memory Phase 0: normalized schema (T01) + Claude adapter (T02) 
- session_memory/core/schema.py: Session/SessionEvent/Cost dataclasses,
  flavor-prefixed uids, watermarks, kind/outcome validation (T01)
- session_memory/adapters/claude.py: JSONL -> Normalized bundle, turn DAG
  via uuid/parentUuid, kind mapping, cost from message.usage (T02)
- tests: schema round-trip + adapter (synthetic + real local session)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-06 19:06:10 +02:00