Seven features from the 2026-10-31 agent-knowledge research, no schema
change (kb_schema_version stays 1.5):
- F1 session-start hook: session_start.py --hook, start_hook.py, opt-in
via session.start_hook in ckb.yaml (D-0011)
- F2 kept answers in outputs/answers/, lint check 16 (D-0013)
- F3 ingest check-in before writing; batch mode skips it
- F5 open threads in workload notes, collected at session start (D-0012)
- F6 dedup_inbox.py: duplicates reported before ingest
- F19 lint check 15 for secrets and <private> spans, --secrets before
every sync commit, redaction in saved conversations (D-0013)
- F20 context budget: lint_report.py --budget, budget in ckb.yaml
Smoke test covers every new script. Docs updated in EN and PL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Four deterministic-script defects plus a contract gap, all reported by a
downstream KB running 1.11.0. None of them crashed; each produced confident,
wrong output, which is why they went unnoticed.
retrieve_candidates.py:
- load_pages() walked rglob unsorted, so tests/expected/retrieve.json encoded
one machine's directory order and the smoke test failed on a clean clone.
- cited_by used setdefault, crediting a shared source to whichever page was
walked first and dropping the source from the results entirely, so the other
citing pages reported corroborated_by: []. Now every citing page is credited.
lint_report.py:
- check 12 reported a cited directory as "does not exist", because is_file()
is false for a directory. Citing a subtree is now explicitly legal and
carries no fingerprint; same fix in lint_quick.sh and lint_quick.ps1.
- check 8 keyed log entries on timestamp + file with the action type dropped,
so a Rule B POINTER was indistinguishable from a duplicate of the entry it
points at. POINTER is now exempt; genuine duplicates still fire.
- check 13 required a backticked path, so "— Chat conversation, Name, date"
counted as unattributed although AGENTS.md lists it as a first-class source.
Attribution and verifiability are now separate questions.
AGENTS.md documents both resolved contract gaps: directory citations, and the
two legal attribution shapes. The fixture KB gained coverage for all three lint
defects — each finding fires under the old script and is silent under the new.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- D-0003: <python> interpreter resolution, per-OS install guide, lint_quick.sh/.ps1
session-start fallback, manual lint checklist, no-script retrieval path,
exporters declare Python a hard requirement
- Lint check 13 pairs each Crux quote block with the attribution under it, so a
multi-source Crux no longer cross-fails; fixture KB gains a second source
- Session summaries for 2026-09-29 and 2026-10-01
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Graft keeps a derived, disposable code graph in sync with a content hash
rather than a calendar, and keeps a protected block on every regenerated
node. This KB is the opposite kind of store — durable, curated, built from
material that cannot be regenerated — but several of Graft's mechanisms
port cleanly, and two of them close real gaps here.
Schema 1.5 is additive: every 1.4 page remains valid.
1. `## Crux` — verbatim source excerpts alongside the synthesis. A summary
can drift silently; a quote either still matches its source or it does
not. Lets `ckb-retrieve` ground an answer without a round-trip to the
archive, and makes drift mechanically detectable.
2. `## Notes` — human-authored and protected everywhere. Closes a real
gap: `ckb-index-external` regenerates connector pages wholesale, so an
annotation written there was previously destroyed on the next refresh.
3. `source_fingerprint`/`source_checked` — a digest of the material a page
was built from. Freshness by date says a page has aged; a fingerprint
says whether its evidence moved. Most valuable for connector-backed
libs, where documents change with no notice.
4. `lint_report.py --quick` — a deterministic one-line session-start
signal, wired into Rule E next to the existing `git status` check.
5. In-degree as a rank-fusion signal in `ckb-retrieve`, weighted below 1.0:
centrality is a prior, not evidence.
6. Blast radius — a new `ckb-ingest` step walking the graph backwards from
touched entities to find what the incoming material contradicts, before
writing anything. Ingest was additive-first, which is how a wiki
accumulates two pages that quietly disagree.
7. Edge vocabulary in `wiki/graph/index.md` rewritten as a question per
verb, and completed: `part_of` was written by `ckb-code-map` but never
declared. Added `produces`, `configures`, `validates`, `implements`.
Lint gains checks 12 (fingerprint drift), 13 (crux verbatimness) and 14
(the protected-Notes rule), verified against a synthetic fixture covering
stale digests, missing sources, fabricated quotes and paraphrased evidence.
Not adopted: the gitignored regenerable store, the MCP server and CLI
daemon, tree-sitter parsing, statusline hooks, telemetry.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Graph edge endpoints use project-root-absolute paths per Rule C
(`/wiki/entities/foo.md`), while the `known` page set is keyed on paths
relative to `wiki/`. check_graph compared them directly, so every
conformant edge was reported as pointing at a missing page. Add the
`/wiki/`-stripped form to the candidate set, matching the normalisation
the decision-record check already performs.
Bump VERSION 1.6.0 -> 1.6.1.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Decision log (VERSION 1.6.0, kb_schema_version 1.4):
- wiki/decisions/ scaffold — numbered NNNN-slug.md records, own index
(with status vocabulary) and log
- type: decision adds status/decided_on/decided_by/affects/review_on;
supersedes/superseded_by carry history and must be set on both sides
- New ckb-decide skill: records decisions and answers what/why/who/when,
what superseded what, and what is still open. Decision pages are
append-only — a changed mind is a new superseding decision
- Graph gains decided_by and affects edge types
- ckb-ingest routes decisions found in raw material to this format;
ckb-retrieve gains the decisions index as a cascade step
Scriptified the mechanical skills:
- ckb-export-okf/scripts/export_okf.py does the whole OKF transform
(frontmatter remap, link rewriting, index/log regeneration, conformance
validation); --check validates without writing
- ckb-lint/scripts/lint_report.py does the read-only detection half
(conformance, freshness, confidence, retention, decisions, orphans,
graph, index/log, source.yaml); judgment calls stay with the model
Also: removed the duplicate personal quiz skill, fixed stale cbk-quiz
doc paths, gitignored __pycache__.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>