It imports claude-code/testing, which only exists inside Claude Code, so the ratchet reported it as an unexpected failing file. It joins the
excluded list (scripts/ci-test-excluded.txt) like every other plugin kit test.
Co-Authored-By: RuFlo <ruv@ruv.net>
`Test Suite (ubuntu-latest)` failed with "71 unexpected failing file(s)": 66 plugin kit tests that import `claude-code/testing` (only
resolvable inside Claude Code) and 5 node:test files vitest finds no suite in. None is a regression; each has its own gate
(`claude plugin test` in every plugin smoke, `node --test` in all-plugins-smoke). They are now listed in scripts/ci-test-excluded.txt and
passed to vitest as excludes. A listed file that no longer exists fails the ratchet, and the baseline stays untouched and shrink-only.
Co-Authored-By: RuFlo <ruv@ruv.net>
Every guard still refuses exactly the same inputs (4752-case A/B against origin/main: 0 decision diffs, 744 message-only diffs).
When music, neural-trader, migrations, observability, rvf or ruvector refuse a memory_store whose namespace they do not own, the
message now says a secret-shaped value was found in a memory write and which namespace was targeted (short plain text, stripped and cut,
dropped if secret-shaped), never an owner and never content. Patch bumps: music 0.2.3, neural-trader 0.3.4, migrations 0.3.3,
observability 0.3.3, rvf 0.3.5, ruvector 0.3.4.
Co-Authored-By: RuFlo <ruv@ruv.net>
Adds scripts/live-ruflo-mods-trim.sh and the measured report: 4,236 fewer prompt tokens with 59 of 69 types hidden; an unnamed hidden type is refused at dispatch; a prompt-named type is not offered in a headless listing.
Co-Authored-By: RuFlo <ruv@ruv.net>
The walker dropped input past its node, character and per-string budgets
silently, so 32 guards passed a secret placed after the cap. textsOf now
appends a TRUNCATED marker; names/hasSecretIn treat it as a finding. Regenerated
the 38 copies, bumped the 39 plugins, emptied the known-holes baseline.
Co-Authored-By: RuFlo <ruv@ruv.net>
Adds structured (name/value pair) and budget (past the node and character
caps) probe classes. Finds 32 guards that pass a secret past the 25000-node
walker budget; recorded in the known-holes baseline, not fixed here (the
root cause is the vendored shared region owned by the sync generator).
Co-Authored-By: RuFlo <ruv@ruv.net>
Extend scripts/sync-mod-screen.mjs to a second kit: the flag() helper in hooks/options.ts of 38 mods now sits between SHARED FLAG markers and is
rewritten/checked from ruflo-agentdb. Drift, a lost marker or zero copies fails --check. Bump the 39 touched plugin versions and their version pins.
Co-Authored-By: RuFlo <ruv@ruv.net>
Run against a real initialised AgentDB with real MCP servers and headless haiku
(v3/docs/validation/agentdb-recall-live-2026-10.md, scripts/live-agentdb-recall.sh).
Defects found by the live run and fixed:
- status.json only flushed on attach/block: skipped, timed-out, cached and error
counters were invisible; add lastReader and lastError
- a permission-refused or failing reader call was swallowed and read as "nothing
relevant"; now counted in errors/lastError
- memory_search (the only semantic reader) was never used; it is now the first
reader, whole prompt only, and its 60-character value cut is completed with
memory_retrieve before the screen and the frame
- ruvector scores arrive as strings and are noise (-0.10..0.10); read numeric
strings and skip results scoring under 0.25
- an all-unsafe answer ended the search, hiding a clean memory behind a poisoned
one; the walk continues and dropped counts each note once
- readers were cached even when none was connected
- add directed-at-the-assistant phrasing rules (plugin-local): haiku followed 6 of
10 injected directives that passed the shared screen, whatever the frame said
41 plugin tests (was 26), smoke 19/19, claude plugin validate passes.
Co-Authored-By: RuFlo <ruv@ruv.net>
Index of ADRs 404-451 with status, builds-on, PR and tests. Corrects ADR-404/444/447
statuses that contradicted shipped code, 444-446 dates (committed 2026-10-04), stale
follow-ups in 446/448, adds superseded-in-part back-links (422, 424, 430 amendments)
and scripts/check-adr-links.mjs (+node:test) for broken links and duplicate numbers.
Co-Authored-By: RuFlo <ruv@ruv.net>
Replaces 37 copy-pasted recursive walkers with a shared breadth-first one (keys as key=value, overlapping windows for long strings, node and char budgets) and widens INVISIBLE (soft hyphen, bidi isolates, Mongolian separator, more). Probe holes 788 -> 0.
Co-Authored-By: RuFlo <ruv@ruv.net>
A baselined guard that stops refusing anything is now a failure, not 26 fixes;
fixed holes need an actual pass. --report preserves hand-written text above the
generated marker. Two tests added; report regenerated.
Co-Authored-By: RuFlo <ruv@ruv.net>
scripts/probe-mod-guards.mjs bundles each hooks/guard.ts with esbuild and runs a
97-probe corpus through its verdict: invisible/bidi splits, nesting to 5000,
wide/huge inputs, secrets in object keys, encodings, regex-DoS stress timed per
call (50 ms gate), benign look-alikes. Calibrates each guard's real secret
surface first so a miss is a real bypass. A fast subset runs last in
smoke-all-plugins against a known-holes baseline. Findings (4 root causes across
32 guards, no ReDoS/crash/hang) are in v3/docs/validation/mod-guard-probe-2026-10.md.
No plugin code changed.
Co-Authored-By: RuFlo <ruv@ruv.net>
Adds scripts/check-guard-tool-names.mjs, an allowlist for tools other servers
serve, node:test tests, and a step in the all-plugins-smoke workflow. Fails
loud when nothing is extracted or the registry is empty.
Co-Authored-By: RuFlo <ruv@ruv.net>
ruv