bchanot/claude
1
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore(tasks): model-router contract criteria 7-11, TODO hardening done + residuals, journal

Browse Source
This commit is contained in:
bchanot
2026-10-08 16:53:43 +02:00
parent 346d6aeab2
commit ae0179f491
3 changed files with 9 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.claude/tasks/TODO.md
+2 -1
View File
@@ -6,7 +6,8 @@ model switch spike-first then flag off; load via `CLAUDE_CODE_PLUGIN_DIRS` + lin
migration of shifters/pins/model-gate in wave 2 after proof; names model-router / route / /route.
- [x] W0 spike in dev-mods (hot reload): facts a-d established 2026-10-08 (plan file § Spike facts); e moved to W1.10
- [x] W1-A the mod in `mods/model-router/` (b721c94, contract `2026-10-08-model-router-w1a-1533`, plan r3): challenge 3 lenses + 1 confirmation (2 BLOCKER + 10 MAJOR closed by named changes), feater DONE first pass, GATE 0 MET 5/5, verifier CONFORME 6/6, security PASS (4 MEDIUM + 5 LOW reported, below)
- [ ] W1-A hardening (security report, user decision): `/route` command origin check (`composer` only); in-agent `route` must not re-model the agent (strip `route.model` for agent loops, drop the dead per-step model path, align the tool description); ReDoS caps on config patterns (length ≤ 200, test ≤ 4 KB); one log per session in every `.catch` + log the silent config drops; phase key charset `^[a-z][a-z0-9_-]{0,31}$`; `additionalProperties: false` on the tool schema + `typeof e.skill`; post-`next` bookkeeping in try/catch; config file size cap 64 KB. Deferred by design: effort ceiling for model-declared routes; window check beyond haiku.
- [x] W1-A hardening (user go "oui durcis", contract criteria 7-11): `/route` composer-only; no agent model axis (effort only after spawn); pattern ≤ 200 / scan ≤ 4096 / phase keys `^[a-z][a-z0-9_-]{0,31}$` / config ≤ 64 KB / `additionalProperties: false` / `typeof e.skill`; `warnOnce` in all 14 catches + config-drop logs; `safely` around post-`next` bookkeeping. feater DONE, gap round (dead `Loop.frozen`, spawn returns `started` verbatim, tool description), GATE 0 MET 9/9, verifier CONFORME 11/11, security PASS.
- [ ] W1-A residuals (security, accepted, none exploitable from outside the user's own files): ReDoS is size-bounded only (`(a+)+$` in `~/.claude/model-router.json` + a 4 KB paste hangs the hook; fix = nested-quantifier rejection or a far lower scan cap); `stat().size` trusted (FIFO/device path in ~/.claude); unrestricted `models`/`agents`/`skills` KEYS echoed raw in logs (log flood); `agentId` read from the flat tool event (engine strip unverified); 5 closures without a kit test (no fs in the kit, LRN-206); "nothing routed" catch text after a state write.
- [ ] W1-B install + docs: `mods/` symlink + `CLAUDE_CODE_PLUGIN_DIRS` in settings.json env via link.sh, root `.gitignore` for the engine-laid `mods/*/tsconfig.json` + `.claude-plugin/types/`, doctor line, README/USAGE/CHANGELOG (doc-sync deferred here from the 1-A /feat run: nothing to document before the install exists), live test in session (swap the spike for the real mod)
- [ ] W2 migration: 15 skills off `Skill(effort-*)`, remove shifters + effort-pins + model-gate, census repointed, docs
- [ ] W3 optional: step heuristics, haiku classifier, quota-aware downgrade, A/B