@event4u/agent-config 10.0.0 → 10.2.0

package/.claude-plugin/marketplace.json

@@ -6,7 +6,7 @@
   },
   "metadata": {
     "description": "Bootstrap shim for the event4u agent-config suite. This plugin ships ONLY the dispatcher hooks plus one install-pointer skill — ALL content (skills, rules, commands) is installed via `npx -y @event4u/agent-config init`, which writes the file projection AND registers the same hooks in ~/.claude/settings.json. DEPRECATED as a content channel; kept so the marketplace listing stays discoverable and existing installs keep working hooks.",
-    "version": "10.0.0",
+    "version": "10.2.0",
     "keywords": [
       "agent-config",
       "skills",

package/CHANGELOG.md

@@ -339,15 +339,177 @@ Entry-shape contract: [`docs/contracts/CHANGELOG-conventions.md`](docs/contracts
 > that forces a new era split (`# Era: 9.37.x`, etc.) — see
 > [`docs/contracts/CHANGELOG-conventions.md § Era splits`](docs/contracts/CHANGELOG-conventions.md).
 
-## [10.0.0](https://github.com/event4u-app/agent-config/compare/9.36.0...10.0.0) (2026-08-12)
+## [10.2.0](https://github.com/event4u-app/agent-config/compare/10.1.0...10.2.0) (2026-08-13)
 
 ### Release highlights
 
 <!-- Curated head: fill before merge, keep it under 10 lines, and leave `_none_` where it is genuinely the answer. The generated log below is unchanged. -->
-- **Behaviour changes:** _auto-derived, rewrite before merge:_ rule/schema diffs, breaking commits or removed public surface in 030ca0d, c1bc2aa, 4eda4ff.
+- **Behaviour changes:** _auto-derived, rewrite before merge:_ rule/schema diffs, breaking commits or removed public surface in 1bb1076, 698aa3f, d805133, 973787a.
+- **Default changes + migration:** _none_
+- **Security and correctness:** _auto-derived, rewrite before merge:_ security-scoped commits in 0309b8e.
+- **Honest nulls:** _auto-derived, rewrite before merge:_ commits carrying an honest-null marker in c428ac6.
+- **Known limitations:** _none_
+
+### Features
+
+* **process:** give operator symptoms a home and codify the loop that resolves them ([6e01a24](https://github.com/event4u-app/agent-config/commit/6e01a24bb9c0a2496feca0a5dfb69d50fefa5fbf))
+* **gates:** warn on unresolved operator symptom entries past 30 days ([ef98cfa](https://github.com/event4u-app/agent-config/commit/ef98cfa7d02c98ec88abf0ec9b778af12b2b6e91))
+* **rules:** size-gated reads — probe a large file before loading it whole ([1bb1076](https://github.com/event4u-app/agent-config/commit/1bb10767269e1d348a874b845f0176fc1c83deba))
+* **hooks:** severity follows the input type — prose alone may not block ([2716d4f](https://github.com/event4u-app/agent-config/commit/2716d4f87280b71f7301a885a9c10da1f1f00ffd))
+* **settings:** gate the deletion side of the surface, not just the addition side ([da3bc3c](https://github.com/event4u-app/agent-config/commit/da3bc3c536184462c73a7214be953b8a57f9c432))
+* **settings:** delete the six keys no code path read ([973787a](https://github.com/event4u-app/agent-config/commit/973787a6026d9b18f86f9961d7286993c6a491b8))
+* **settings:** add the disposition axis to lint_settings_classes ([4c9ced8](https://github.com/event4u-app/agent-config/commit/4c9ced8908a9e89e2f9b8684cb4f1f6642d55c55))
+* **conformance:** give task-completeness a denominator, and cover it ([18e4a0c](https://github.com/event4u-app/agent-config/commit/18e4a0c85cd705e5f404ea5b08f7a1018b7a7c3b))
+* **conformance:** measure task completeness before refusing on it ([11610f8](https://github.com/event4u-app/agent-config/commit/11610f85c46d0eeb024ccb224e9236d264ffe0fe))
+
+### Bug Fixes
+
+* **roadmap-writing:** fit the symptom-loop route without tripping the size check ([d5b3253](https://github.com/event4u-app/agent-config/commit/d5b3253f7875ab4da2da45a869cbbcb134a03763))
+* **roadmaps:** anonymize the vendored corpus upstream outside its attribution file ([8406ac8](https://github.com/event4u-app/agent-config/commit/8406ac8c52a75355e0c704184aea123c7895d2fc))
+* **roadmaps:** satisfy the complexity and blocker contracts on two active plans ([498cea7](https://github.com/event4u-app/agent-config/commit/498cea7c93bdbea57e00b5d6ba56ca828692e34f))
+* **contracts:** the rule-router described a Claude Code gap that closed in PR #1231 ([f87240d](https://github.com/event4u-app/agent-config/commit/f87240d899e5418fa256e6462160b74466ad728d))
+* **context-hygiene:** name the wait shape both loop clauses miss ([698aa3f](https://github.com/event4u-app/agent-config/commit/698aa3f65e704e3fa796be606c54a880c0736d01))
+* **turn-end-gate:** stop refusing the honest "not done yet" line ([f85a8c7](https://github.com/event4u-app/agent-config/commit/f85a8c713d7bfa1d577ed1897261493ab1336d33))
+* **git-guard:** classify the verb that is invoked, not the one that is named ([1b0d42d](https://github.com/event4u-app/agent-config/commit/1b0d42dd4135af012647853a469b1e43070d6502))
+* **transcripts:** close the store-slug character class instead of widening it ([580f12d](https://github.com/event4u-app/agent-config/commit/580f12ddb9e1dc1c2d0a5ca9d0d2961707fa3b76))
+* **evidence:** a worktree location is not a verdict subject ([351f245](https://github.com/event4u-app/agent-config/commit/351f245b946d5ec186d454888dfa65764003de32))
+* **review:** stop the findings artefact citing a path that is the finding ([3dc1f86](https://github.com/event4u-app/agent-config/commit/3dc1f863d6c15b1afdccf0700897cc463eb80ccb))
+* **council:** close the seven review findings, including two tests of mine that could not fail ([368c895](https://github.com/event4u-app/agent-config/commit/368c895d78d50b79520b686ade54e7d9aeddfe76))
+* **council:** the spawned member answered the caller's hooks, and held tools it never needed ([67efa72](https://github.com/event4u-app/agent-config/commit/67efa720e380e54989c0e95b14285b729f5e279b))
+* **council:** gemini rejects --system too, and two members dropped the system prompt entirely ([9faee19](https://github.com/event4u-app/agent-config/commit/9faee19e22fcf51cd21c04b7b12bd0a4da7bfb4e))
+* **council:** close six of the eight review findings ([01c7b1f](https://github.com/event4u-app/agent-config/commit/01c7b1fb23ecfee991477d5c086ebc977fa5720b))
+* **roadmap:** make the PR the end condition of /roadmap:next, not a step ([0184f3b](https://github.com/event4u-app/agent-config/commit/0184f3b92c8ce975c6f9eab815060f09e0479149))
+* **council:** print the post-run quorum, not only the pre-run one ([cd8e518](https://github.com/event4u-app/agent-config/commit/cd8e518c115b2b873e2c85fc8aaf1c85edc37f67))
+* **council:** codex exec has no --system flag, so stop passing one ([990d698](https://github.com/event4u-app/agent-config/commit/990d6982ff7f79c1979a96e5b8a37beaa788e65c))
+* **review:** close the R2 findings, one of them by admitting the fix was half-built ([ff8be31](https://github.com/event4u-app/agent-config/commit/ff8be31c5aea39a2b85874b1d1224f463a12b0d7))
+* **session-eol:** stop recommending /clear without checking the envelope exists ([ee27132](https://github.com/event4u-app/agent-config/commit/ee2713277230081929096719893739f64b416913))
+* **session-recycle:** refuse a write into an unanchored working directory ([825df09](https://github.com/event4u-app/agent-config/commit/825df094771c44bbab37de44ea03c713e39980cd))
+
+### Documentation
+
+* **roadmaps:** archive the symptom-driven harvest loop, fully executed ([d8ddccb](https://github.com/event4u-app/agent-config/commit/d8ddccb8707be0e39b15106f5181964c6f32f531))
+* **roadmaps:** adopt the four August plans plus their program layer ([f84ba70](https://github.com/event4u-app/agent-config/commit/f84ba70e4bc56c9413d05ae6f9aa0189c3e629d1))
+* **review:** re-bind the skip declaration to the Phase 2 re-cut scope ([ceac026](https://github.com/event4u-app/agent-config/commit/ceac02600bc198309163b7d269b87c6c648e4049))
+* **roadmap:** re-cut Phase 2 to the downstream evidence control, cancel Phase 3 ([614f3df](https://github.com/event4u-app/agent-config/commit/614f3dfdfe6f4eb3e0c805769bd19ad1846c61ed))
+* **contracts:** correct the prompt-channel residual — the prompt IS readable ([35a7796](https://github.com/event4u-app/agent-config/commit/35a7796ce728c6c6e70d10d813eb0da083c343bb))
+* **roadmap:** corpus knowledge skills — the compiler gap, with eight corrections to the source ([c428ac6](https://github.com/event4u-app/agent-config/commit/c428ac61f8da7aa8e83f0f71758166fcf7e2eb42))
+* **review:** declare the structured-guard-input measurement a no-code-surface skip ([44c5092](https://github.com/event4u-app/agent-config/commit/44c5092c522fbf4cff0db2ade5e731266c521784))
+* **evidence:** measure the three Phase-1 falsifiers before building the field ([e817d50](https://github.com/event4u-app/agent-config/commit/e817d5084d67d5d937e65fcf86926ab4e640a566))
+* **contracts:** state which text a guard reads, pre- or post-expansion ([953e7c1](https://github.com/event4u-app/agent-config/commit/953e7c1efca019061ba37842726fcf8fd27a198e))
+* **review:** declare the waiter-discipline completion a no-code-surface skip ([cf2019e](https://github.com/event4u-app/agent-config/commit/cf2019ea2da0473ee36906f99b98a76be50541fa))
+* **roadmap:** regenerate the dashboard after resolving the merge ([f2e6179](https://github.com/event4u-app/agent-config/commit/f2e617964b360f61d3cf32cf3094564498f25f44))
+* **roadmap:** risk register for the structured-guard-input plan ([224eda9](https://github.com/event4u-app/agent-config/commit/224eda90be5f1918bc0f59cf6aa0072ef0c259a1))
+* **council:** record the convergence and the work it does not fit in ([806e40c](https://github.com/event4u-app/agent-config/commit/806e40c25dbc148d51cccd51438c6622c451f9bc))
+* **roadmap:** close zero-settings 2.1, 2.2 and 4.1, and re-date the blocker ([9acf679](https://github.com/event4u-app/agent-config/commit/9acf679f83a5ec260cc0b8be73aa98cd5bb3a219))
+* **roadmap:** regenerate the dashboard after the merge ([945d614](https://github.com/event4u-app/agent-config/commit/945d6148a4e1cda1f01bdd8da769b58bd66a880e))
+* **roadmap:** record that a failed council attempt still spends quota ([d6e6f11](https://github.com/event4u-app/agent-config/commit/d6e6f11ab78b2ec593098b3e850be94e7e342800))
+* **roadmap:** close zero-settings 2.1, 2.2 and 4.1; record why 3.1 and 3.2 stay open ([c5499d9](https://github.com/event4u-app/agent-config/commit/c5499d94529da815709afa2d9776b69acc6ee5b4))
+* **settings:** the telegraph rule states its own scope, and four contracts stop naming a dead key ([d805133](https://github.com/event4u-app/agent-config/commit/d805133257302e0beaa732cb1dd33ee413ceb4a6))
+* **settings:** republish the counts, the floor, and the reclassification argument ([e53a799](https://github.com/event4u-app/agent-config/commit/e53a7992f4eb49b0859a267c7814434a263fdf4e))
+* **audit:** record the cross-project session audit and its refusals ([dddafe9](https://github.com/event4u-app/agent-config/commit/dddafe91e36f2eceb243d8292e1da087eeff00d0))
+* **review:** re-bind the adapter-follow-up artefact to the fixed scope ([783b784](https://github.com/event4u-app/agent-config/commit/783b784456429ff897538f23ffd83250c8e67b89))
+* **review:** record the R2 findings on the adapter follow-up before fixing them ([a910931](https://github.com/event4u-app/agent-config/commit/a910931b3fe915c97902b90ecdea9540d4b07284))
+* **review:** re-bind the context manifest, not only the marker ([3a22d35](https://github.com/event4u-app/agent-config/commit/3a22d357e7d6f5c93e4a8b855a58dd423762c09f))
+* **review:** re-bind the R2 artefact to the fixed scope ([e6b4ce8](https://github.com/event4u-app/agent-config/commit/e6b4ce8e968d54a5bb22ca0b8d9f0d36ee7069fb))
+* **review:** record the R2 completion review before the fixes ([169fcfd](https://github.com/event4u-app/agent-config/commit/169fcfd9d454ef9951c698783ea8859ff94462a4))
+* **evidence:** task-completeness measures 0 of 3 precision — detector D is not built ([a7f6ffd](https://github.com/event4u-app/agent-config/commit/a7f6ffd459c7b5f19db087b31ccd42de5a85504b))
+* **roadmap:** close zero-settings Phase 1 + 4.2, file the consent blocker ([99e6df2](https://github.com/event4u-app/agent-config/commit/99e6df2aebfae9433eff3ba7ce24c46090706196))
+* **settings:** classify all 140 leaves and ratchet the derivable surface ([4a40578](https://github.com/event4u-app/agent-config/commit/4a405780354a7432d615373577f2e2f0bba4efa6))
+* **review:** ship the reviewer prompt and scope beside the verdict ([0e57c8e](https://github.com/event4u-app/agent-config/commit/0e57c8e27621350fe086304b6f8f4f9143ffb8f4))
+* **review:** re-bind the R2 artefact to the fixed scope, dispositions terminal ([8492864](https://github.com/event4u-app/agent-config/commit/8492864e2a6540fad5d1970cf76ae6a0dea2259c))
+* **evidence:** pre-register the bar for detector D, contamination disclosed ([ee3f43f](https://github.com/event4u-app/agent-config/commit/ee3f43faca834a44eaaf3dab1770e9f422aeae9c))
+* **review:** record the R2 findings before fixing any of them ([a151f36](https://github.com/event4u-app/agent-config/commit/a151f368422afbd18cf1f81175e69129acfa0e05))
+* **roadmap:** close the rootless-write roadmap and record the sibling inventory ([4f5447a](https://github.com/event4u-app/agent-config/commit/4f5447a5ce022adf8e356ebc93419cababa1af0a))
+
+### Refactoring
+
+* **settings:** delete the five keys nothing ever read ([94b9134](https://github.com/event4u-app/agent-config/commit/94b913481610deb1e981760280019c7ef0ae4aa1))
+
+### Tests
+
+* **settings:** pin what a deleted unread key can and cannot do ([db0c6bd](https://github.com/event4u-app/agent-config/commit/db0c6bdba4eef271ae6803e4adb12c202d14ee2b))
+* **settings:** pin the inverted invariant per deleted key ([e87af34](https://github.com/event4u-app/agent-config/commit/e87af347d9f0eacfd5f753737fadc85813b181fb))
+
+### Build
+
+* take the trunk install bundle instead of a worktree rebuild ([f3ef74c](https://github.com/event4u-app/agent-config/commit/f3ef74c800b492cab4327953de7bb8af706e83bc))
+* regenerate the derived outputs after the reconciliation merge ([50b4ce6](https://github.com/event4u-app/agent-config/commit/50b4ce6df13dff915ec0adbbe4f03f0f08475d02))
+* rebuild the tracked install bundle after the schema deletion ([7b143ae](https://github.com/event4u-app/agent-config/commit/7b143ae4e0611f478e0ddd86c8e837d01a738940))
+
+### Chores
+
+* **index:** regenerate the artifact index and catalog for the new guideline ([5d2d563](https://github.com/event4u-app/agent-config/commit/5d2d56351048163930fba8c8871d4eaddb5d203b))
+* **security:** backstop the anonymization obligation for one more harvest source ([0309b8e](https://github.com/event4u-app/agent-config/commit/0309b8efc7f304bdbbc1886701dad1706c22f543))
+* **tests:** drop the template handle the trimmed assertion orphaned ([ed764a9](https://github.com/event4u-app/agent-config/commit/ed764a9e600d330e5a33d3c49981f2e7e84b840e))
+* **tests:** finish the orphan chain the previous deletion opened ([9bc6020](https://github.com/event4u-app/agent-config/commit/9bc6020a9ddf9578b98898024a2191ffc480a906))
+* **tests:** drop two dead declarations the changed-files linter surfaced ([2d7ff19](https://github.com/event4u-app/agent-config/commit/2d7ff19b3dba4584b83e393a5e6f6c2fdafdd816))
+* **roadmap:** archive road-to-feedback-9-35, which merged complete but unarchived ([2e26094](https://github.com/event4u-app/agent-config/commit/2e26094de27c845466229c6e7842a8217ac3b21f))
+* **roadmaps:** archive road-to-completion-loop, closed as a published null ([92f9b9a](https://github.com/event4u-app/agent-config/commit/92f9b9aa675cbc42cc81e7eb829e68d6ddd3c4b7))
+
+Tests: 13506 (+95 since 10.1.0)
+
+## [10.1.0](https://github.com/event4u-app/agent-config/compare/10.0.0...10.1.0) (2026-08-12)
+
+### Release highlights
+
+<!-- Curated head: fill before merge, keep it under 10 lines, and leave `_none_` where it is genuinely the answer. The generated log below is unchanged. -->
+- **Behaviour changes:** _auto-derived, rewrite before merge:_ rule/schema diffs, breaking commits or removed public surface in 0992c92, 20a8606, 63aeba1.
 - **Default changes + migration:** _none_
 - **Security and correctness:** _none_
-- **Honest nulls:** _auto-derived, rewrite before merge:_ commits carrying an honest-null marker in 7c6b404.
+- **Honest nulls:** _auto-derived, rewrite before merge:_ commits carrying an honest-null marker in 0f3ce9b.
+- **Known limitations:** _none_
+
+### Features
+
+* **untrusted-content:** tag external content where it enters, with an unforgeable boundary ([7a313ec](https://github.com/event4u-app/agent-config/commit/7a313ecfae4eadd06ff547a4a67d363a6758707b))
+* **skill-schema:** structured runtime requirements, and the rule that gives them teeth ([63aeba1](https://github.com/event4u-app/agent-config/commit/63aeba1070af3d72bcaac0c7f999788d3b833354))
+
+### Bug Fixes
+
+* **roadmap,context:** satisfy the two gates this branch newly tripped ([e61d461](https://github.com/event4u-app/agent-config/commit/e61d46142a9dc41841c98360377cc89176ce1057))
+* **roadmap-archival:** an open blocker outlives its steps, and the sweep did not look ([710dc26](https://github.com/event4u-app/agent-config/commit/710dc266a27a82aad9a527c525650e8d43accfdc))
+* **review:** close all eight R2 findings, one of them by admitting a gap instead of papering it ([0992c92](https://github.com/event4u-app/agent-config/commit/0992c92219364bc646e78c29fa3072b5e46361ef))
+* **diff:** drop a regenerated report my git add -A swept in ([3aec031](https://github.com/event4u-app/agent-config/commit/3aec031b160a8bc676053be370b2fddfbeab250c))
+* **skill-schema:** rename requires to runtime_requires — the key was already taken ([20a8606](https://github.com/event4u-app/agent-config/commit/20a86065cc5d9c86e9fdf0710c51ddf850ca7ae4))
+* **roadmap:** anonymize the harvest source per source-confidentiality ([195e444](https://github.com/event4u-app/agent-config/commit/195e444a7b1f9d7d05ac5cb768b0be17a1ac58ea))
+
+### Documentation
+
+* **changelog:** curate the 10.0.0 head, which shipped the same placeholder hours later ([0f3ce9b](https://github.com/event4u-app/agent-config/commit/0f3ce9b22c1858e12556a4b0fe95bec1931dde91))
+* **roadmap:** the verified residue of five external release reviews ([a25579f](https://github.com/event4u-app/agent-config/commit/a25579fdf4d13cf6057ccdfeba5dc87990dd11e0))
+* **analysis:** threat-model the confirmation store, and enumerate the buried decisions ([8b18f1c](https://github.com/event4u-app/agent-config/commit/8b18f1c4a850a9b08906aceccbe3562f381de7c3))
+* **records:** three figures five reviews keep re-deriving, corrected at the source ([1612a70](https://github.com/event4u-app/agent-config/commit/1612a70e3e13daa0572fefb4803533708c1273f0))
+* **changelog:** curate the 9.36.0 head that shipped its own placeholder ([83a3922](https://github.com/event4u-app/agent-config/commit/83a3922f63efbe5aca280e775acc85b36573c3f4))
+* **review:** re-bind after the count correction ([9dd75ec](https://github.com/event4u-app/agent-config/commit/9dd75ecf178152641f8744a8f1b81f045befde7b))
+* **roadmap:** correct the test count the review fixes moved ([ec52c42](https://github.com/event4u-app/agent-config/commit/ec52c421ba69dc85ee78095323f8495b93f3dd26))
+* **review:** re-bind the R2 artefact to the fixed scope, dispositions terminal ([e601648](https://github.com/event4u-app/agent-config/commit/e6016487e883f611b2b7b989f23ed682796a6a73))
+* **review:** record the R2 findings before fixing any of them ([1cf8d78](https://github.com/event4u-app/agent-config/commit/1cf8d787c015f51261bdd22b7388e47d127003f6))
+* **roadmap:** add the Acceptance Criteria the completion review needs to bind to ([f35becd](https://github.com/event4u-app/agent-config/commit/f35becd885c09a5ac7f5a678adf29a360e776efa))
+* **roadmap:** follow the schema key rename, and record why the obvious name was unavailable ([5720847](https://github.com/event4u-app/agent-config/commit/572084763fef97eb66fa91fd722d75a9c527440a))
+* **roadmap:** the executable-payload harvest, with four source claims corrected ([2f4d8a8](https://github.com/event4u-app/agent-config/commit/2f4d8a8be5891d6045d4605a4a6d8fe3dc28a65e))
+
+### Tests
+
+* **council-cli:** the quota case asserted a property of the developer machine ([35037f9](https://github.com/event4u-app/agent-config/commit/35037f96d3e631d6ee1a235415e716e2a29f8f8c))
+
+Tests: 13411 (+27 since 10.0.0)
+
+## [10.0.0](https://github.com/event4u-app/agent-config/compare/9.36.0...10.0.0) (2026-08-12)
+
+### Release highlights
+
+- **Behaviour changes:** council transport is **resolved per machine rather than
+  configured** — the breaking change this major carries (4eda4ff); every member
+  now resolves as `auto`, so a configured CLI call budget applies wherever a
+  council config exists. The turn-end gate is armed unconditionally and loses its
+  settings surface (42cd613). Conformance round 7 records its downgrades rather
+  than restating them (030ca0d, c1bc2aa).
+- **Default changes + migration:** _none_
+- **Security and correctness:** _none_
+- **Honest nulls:** a blocker whose resolution condition no passage of time could
+  satisfy — a default-OFF gate cannot soak, so the switch removal discharges it
+  and the original wording stays verbatim beside the resolution, because the
+  unsatisfiable condition is itself the finding (7c6b404).
 - **Known limitations:** _none_
 
 ### BREAKING CHANGES
@@ -400,8 +562,13 @@ Tests: 13384 (+51 since 9.36.0)
 
 ### Release highlights
 
-<!-- Curated head: fill before merge, keep it under 10 lines, and leave `_none_` where it is genuinely the answer. The generated log below is unchanged. -->
-- **Behaviour changes:** _auto-derived, rewrite before merge:_ rule/schema diffs, breaking commits or removed public surface in 9f69017, 924cad8, 3c20d47, 72bb1bc.
+- **Behaviour changes:** the demand gate reads `project.audience` and is inert at
+  `self` — a market-demand ladder no longer fires on a project that intends no
+  market (9f69017). The two UI rules stop claiming the `ui-route-nudge` concern
+  reads their `keyword:` triggers; it does not, and six surfaces said it did
+  (3c20d47). The design skills now reach the consumers they were written for
+  (72bb1bc). The CI-side change in 924cad8 is internal to the pipeline and
+  changes no consumer-visible behaviour.
 - **Default changes + migration:** _none_
 - **Security and correctness:** _none_
 - **Honest nulls:** _none_

package/README.md

@@ -4,7 +4,7 @@
 
 [![Smoke](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml) [![Public install smoke (3 OS × 2 Node)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml) [![npm](https://img.shields.io/npm/v/@event4u/agent-config?style=flat-square&label=npm&color=orange)](https://www.npmjs.com/package/@event4u/agent-config) [![agent-config MCP server](https://glama.ai/mcp/servers/event4u-app/agent-config/badges/score.svg)](https://glama.ai/mcp/servers/event4u-app/agent-config) [![MCP Toplist](https://mcptoplist.com/badge/glama%2Fevent4u-app%2Fagent-config.svg)](https://mcptoplist.com/server/glama%2Fevent4u-app%2Fagent-config)
 
-[![Skills](https://img.shields.io/badge/Skills-289-orange?style=flat-square)](dist/agent-src/skills/) [![Rules](https://img.shields.io/badge/Rules-116-orange?style=flat-square)](dist/agent-src/rules/) [![Commands](https://img.shields.io/badge/Commands-196-orange?style=flat-square)](dist/agent-src/commands/) [![Guidelines](https://img.shields.io/badge/Guidelines-104-orange?style=flat-square)](docs/guidelines/) [![Personas](https://img.shields.io/badge/Personas-29-orange?style=flat-square)](dist/agent-src/personas/) [![Advisors](https://img.shields.io/badge/Advisors-5-orange?style=flat-square)](dist/agent-src/personas/advisors/)
+[![Skills](https://img.shields.io/badge/Skills-289-orange?style=flat-square)](dist/agent-src/skills/) [![Rules](https://img.shields.io/badge/Rules-116-orange?style=flat-square)](dist/agent-src/rules/) [![Commands](https://img.shields.io/badge/Commands-196-orange?style=flat-square)](dist/agent-src/commands/) [![Guidelines](https://img.shields.io/badge/Guidelines-105-orange?style=flat-square)](docs/guidelines/) [![Personas](https://img.shields.io/badge/Personas-29-orange?style=flat-square)](dist/agent-src/personas/) [![Advisors](https://img.shields.io/badge/Advisors-5-orange?style=flat-square)](dist/agent-src/personas/advisors/)
 
 **Try it in 30 seconds** — drop one read-only subagent into any repo and watch it gate "done": `@production-validator check this branch is actually done`. No wizard, no lock-in, nothing else installed — the [30-second wedge ↓](#quickstart) is the whole first step. Start at the proof, not the catalog: **[event4u-app.github.io/agent-config/proof/](https://event4u-app.github.io/agent-config/proof/)**.
 

package/dist/agent-src/commands/roadmap/next.md

@@ -224,6 +224,11 @@ Chunked commits (the agent picks the split — never ask how, per
 [`/create-pr`](../../git/pr/create.md): archival sweep, preflight,
 push, PR opened **ready for review** (not draft).
 
+This step is where the run **ends**, not a step it may stop before — see the
+second Iron Law below. The invocation already authorised the push and the PR;
+asking for them again is the re-ask
+[`no-cheap-questions`](../../rules/no-cheap-questions.md) § IL 5 forbids.
+
 ### 7. CI — fix locally, honour the push gate
 
 Follow [`/create-pr` § 4d](../../git/pr/create.md): wait for the
@@ -258,6 +263,39 @@ A ROADMAP CLAIMED "IN FLIGHT", "MERGED", OR "DONE" WITHOUT A LIVE CHECK
 AT THE MOMENT OF THE CLAIM IS AN UNVERIFIED CLAIM.
 ```
 
+## Iron Law — the run ends at a PR, never at an offer to open one
+
+```
+STEP 6 IS AN END CONDITION, NOT A STEP THE RUN MAY STOP BEFORE REACHING.
+A RUN THAT PRODUCED COMMITS ENDS WITH A PR OPENED FOR REVIEW.
+THE INVOCATION IS THE ACCEPT — IT COVERS THE BRANCH, THE CHUNKED COMMITS,
+THE FIRST PUSH TO THAT BRANCH, AND THE PR. NEVER RE-ASK FOR THEM.
+"SAY THE WORD AND I'LL OPEN IT" IS A PROMISSORY CLOSE, WHICH
+verify-before-complete ALREADY FORBIDS. SO IS ENDING ON COMMITTED-BUT-UNPUSHED
+WORK WITHOUT NAMING IT.
+STOPPING EARLIER IS LEGAL ONLY AS ONE OF process-full's FIVE NAMED HALTS,
+A HARD-FLOOR TRIGGER, OR AN N=3 EXHAUSTION — AND THE HALT IS NAMED IN THE
+REPORT TOGETHER WITH WHAT IS STILL UNPUSHED.
+```
+
+**What this does NOT lift.** § 7's re-push restriction stands unchanged: the
+invocation covers the *first* push that opens the PR, never the pushes that
+follow a CI fix. Those are Hard-Floor triggers needing the user's word that turn
+([`non-destructive-by-default`](../../rules/non-destructive-by-default.md)), and
+no reading of this law manufactures a per-PR standing mandate.
+
+**Why it needed saying, measured on this command's own run.** A session screened
+30 roadmaps, recovered from a live claim collision, picked, branched and did the
+work — then closed with a status report and no PR, because § 6 read as a step in
+a list rather than as the condition the run terminates on. The user had to ask
+for the PR the invocation had already authorised. A step that can be narrated
+instead of executed is not an obligation; an end condition is.
+
+**Honest scope.** `enforced_by: none` — nothing inspects a chat turn for a
+missing PR, so this is model-carried like the live screen above it. The
+deterministic half is elsewhere and unchanged: `/create-pr` runs the archival
+sweep and preflight, and CI gates the branch once the PR exists.
+
 ## Rules
 
 - **One roadmap per invocation.** Never chain a second pick after the

package/dist/agent-src/contexts/communication/rules-auto/token-efficiency-mechanics.md

@@ -83,10 +83,31 @@ When `personal.minimal_output: true`:
 ### Minimize tool calls
 
 - Parallel reads — don't read 5 files sequentially.
-- Regex search over full file reads. View specific line ranges.
 - One codebase search call with all symbols — not 5 separate.
 - Short question → short answer. Summary tables only for 3+ items.
 
+### Size-gated read procedure (the 800-line threshold)
+
+The rule states the obligation; this is how to discharge it. Three steps, in
+order, and each one narrows the next:
+
+1. **Size** — `wc -l <file>`. Under the threshold, stop here and read it.
+2. **Shape** — a structural grep for the anchors you need, emitting offsets:
+   `grep -nE '^(class|function|def|export|## )' <file>` for a map, or
+   `grep -n '<the symbol>' <file>` when you know what you are after.
+3. **Slice** — a ranged read around those offsets only (the host's
+   `offset`/`limit` read, or `sed -n 'A,Bp'`), widened once if the slice
+   straddles a boundary.
+
+Two failure modes this replaces. **Blind full read:** loading a 3,000-line file
+to answer a question one function answers — the cost is paid on every later turn
+that carries it. **Blind slice:** reading lines 1–200 of an unknown file and
+guessing the rest, which is the probe skipped rather than performed.
+
+The step-2 grep is not a second probe of the same file for the same fact — it
+returns information step 1 did not have, so it is one operation with step 3, per
+the enumerated-set carve-out.
+
 ### Exceptions
 
 - Small output (< 30 lines) — read directly.

package/dist/agent-src/contexts/contracts/frugality-charter.md

@@ -38,7 +38,7 @@ Live Phase 1 schema (see [`templates/agent-settings.md`](../../templates/agent-s
 | `verbosity.offer_council_in_delivery` | When `false`, delivery commands skip the optional "run Council on this?" offer. |
 | `verbosity.post_action_reports` | `off` / `minimal` / `full` — multi-line status blocks after a successful action. |
 | `verbosity.intent_announcements` | When `false`, writers must not generate "Let me…", "Now I will…", "Found it" openers. |
-| `telegraph.speak_scope` | `off` / `prose_only` / `aggressive` — telegraph-speak grammar scope. Iron-Law fenced blocks never touched. |
+| `telegraph.speak` | When `false` (the shipped default) the telegraph-speak rule does not ship at all; when `true` its own carve-out list states the grammar's scope — options blocks, Iron-Law literals, code, and paths stay byte-stable. |
 
 Phase 10 adds `verbosity.script_output` (`silent` / `minimal` / `verbose`) and `verbosity.taskfile_command_echo` (`true` / `false`); writer artifacts cite this row when those land.
 

package/dist/agent-src/rules/context-hygiene.md

@@ -72,6 +72,45 @@ Calling the **same tool** more than **2 times in a row** with similar parameters
 **"Similar parameters" is the load-bearing word — an enumerated set is not a loop.** Walking N *declared, distinct* targets (a downstream-caller sweep per [`downstream-changes`](downstream-changes.md), an override chain, the members of a grep result) is one operation whose parameters differ every call. Counting it as N repetitions makes `downstream-changes`' "find **ALL** callers, tests, imports" unsatisfiable — it cannot be done in two calls. The loop this detects is *repetition without new information*: same target, same parameters, hoping for a different answer. Mirrors the same carve-out in [`token-efficiency`](token-efficiency.md).
 `sequentialthinking` is especially prone to loops: at most **once** per task, NEVER for simple file operations, command execution, or straightforward edits.
 
+### Waiting is one waiter, never a fleet
+
+```
+ONE WAITER PER CONDITION. NEVER A TIMER AND A WATCHER FOR THE SAME WAIT.
+NEVER START A NEW WAITER WHILE ONE IS STILL LIVE FOR THAT CONDITION.
+A NUMBER THAT CHANGED IS NOT NEW INFORMATION UNLESS IT CHANGES WHAT YOU DO NEXT.
+```
+
+Waiting on something external — CI, a deploy, a queue, a background job — is the
+one shape the loop test above systematically misses, in both of its clauses.
+
+- **"Same parameters" fails.** A `sleep 240` followed by `sleep 420` followed by
+  `sleep 595` are three different calls by any parameter comparison, so nothing
+  reads them as repetition.
+- **"Repetition without new information" fails too**, and this is the subtler
+  half: the poll genuinely returns a different value each round — 31 green, then
+  34, then 36 — which *feels* like progress. It is not information unless a
+  different digit would make you act differently. It would not: the only actions
+  available were wait, or stop and report, and 34 licensed neither more than 31
+  did. **The discriminator is the decision, not the digits.**
+
+So state it separately: pick **one** waiter for the condition and let it finish.
+Where the harness re-invokes on completion, waiting costs nothing and polling
+costs a turn per expiry — and every stacked waiter is a turn that arrives *after*
+the answer already did. Where it does not, one `until <condition>` loop is still
+one call rather than N.
+
+**Measured once, at n=1, and worth naming for its shape rather than its
+frequency** (2026-08-12): a CI wait started roughly 35 background commands —
+a fresh timer *and* a fresh condition-watcher per round, none cancelled when the
+next pair started. CI settled with about fifteen still live; each then expired
+and produced an empty turn. Nothing was corrupted, which is exactly why it ran
+so long: every individual step looked reasonable.
+
+**Enforcement: none, and the reason is structural.** The `context-hygiene` hook
+counts tool calls; a waiter is indistinguishable from any other call at that
+layer, and nothing in the envelope says two live waiters are watching one
+condition. This is model-carried on every host.
+
 ## Read-Loop Detection — the 15 / 25 rule
 
 ```

package/dist/agent-src/rules/token-budget-discipline.md

@@ -74,8 +74,9 @@ from. Where a consumer has enabled it, it **exempts** skills marked
 - Preserve all worked examples, decision tables, and code blocks verbatim
 
 This exemption is limited to the skill's guidance content; reply prose from the
-*agent* (not from the skill body) remains subject to telegraph-speak unless the
-user has set `telegraph.speak_scope: off`.
+*agent* (not from the skill body) remains subject to telegraph-speak wherever
+that rule ships at all (`telegraph.speak`), within the carve-outs the rule
+itself names.
 
 ## Value-over-budget escalation
 

package/dist/agent-src/rules/token-efficiency.md

@@ -52,6 +52,28 @@ cohorts (batch), or a cost-aware model/effort decision → route via the
 [`token-optimizer`](../skills/token-optimizer/SKILL.md) index branch
 (`api-cost-levers` row) — single source of truth for the billing levers.
 
+## Size-gated reads — probe a large file before loading it whole
+
+```
+A FILE OVER THE THRESHOLD IS PROBED, NEVER LOADED WHOLE ON SPEC.
+SIZE CHECK → STRUCTURAL GREP → BOUNDED SLICE. FULL READ BELOW IT.
+```
+
+**Threshold: 800 lines.** Above it, establish the file's shape before its
+content — size, then a structural grep for the offsets that matter, then a
+ranged read of those offsets. Below it, just read it: a probe you did not need
+costs a call and teaches nothing.
+
+The number is a **stated default, not a measured optimum** — said plainly rather
+than implying a derivation it does not have. *Revisit-if:* a run records a
+probe-then-slice that cost more than the full read would have, or the host's
+ranged-read primitive changes its own default. Either falsifies the number, not
+the obligation.
+
+Additive: the repetition discriminator above is untouched, an enumerated
+multi-file sweep stays ONE operation, and re-probing a file whose shape you
+already established this session is the loop, not the fix.
+
 ## Mechanics
 
 Anti-loop patterns, act-skip-narration / stop-early / minimize-tool-calls clauses, and small-output / debugging exceptions: [`token-efficiency-mechanics`](../contexts/communication/rules-auto/token-efficiency-mechanics.md). Precedence: never overrides `user-interaction` or command rules.

package/dist/agent-src/scripts/archive_completed_roadmaps.ts

@@ -335,6 +335,23 @@ function archive_completed(
         if (stats.open_ !== 0 || stats.deferred !== 0) {
             continue; // not complete
         }
+        // An unresolved blocker outlives its steps. Closing every box does not
+        // answer a question the roadmap raised for a human, and archiving on
+        // step-count alone is how `b-highlights-mechanism` left the active tree
+        // unanswered in the 9.29 roadmap — four releases before the failure it
+        // predicted shipped in the 9.36.0 changelog head. Same discipline
+        // `roadmap-progress-sync` Iron Law 3 applies to `[~]` steps: a roadmap
+        // carrying an open decision stays visible until the decision is made.
+        const open_blockers = stats.open_blockers;
+        if (open_blockers.length > 0) {
+            const ids = open_blockers.map((b) => b.id).join(', ');
+            process.stderr.write(
+                `  ⚠️  ${stats.rel}: all steps closed but ${open_blockers.length} ` +
+                    `blocker(s) still open (${ids}) — not archived. Resolve them, ` +
+                    'or record the decision, then re-run the sweep.\n',
+            );
+            continue;
+        }
         const old_rel = `agents/roadmaps/${stats.rel}`;
         if (changed_only && !(touched as Set<string>).has(old_rel)) {
             continue; // complete, but not this branch's work

package/dist/agent-src/skills/adr-create/SKILL.md

@@ -11,6 +11,11 @@ execution:
   command:
     - ./scripts-run
     - src/scripts/adr/regenerate_index
+runtime_requires:
+  bins:
+    - bash
+    - node
+  network: []
 workspaces:
   - agent-config-maintainer
 packs:

package/dist/agent-src/skills/ai-council/SKILL.md

@@ -228,10 +228,12 @@ Three vendor CLIs are wired:
       model: claude-sonnet-4-5
   ```
 
-- **OpenAI / Codex** — invokes `codex exec --json` and walks the
-  newline-delimited JSON event stream, pulling text from
-  `item.completed` and tokens from `turn.completed`. Session id
-  is preserved.
+- **OpenAI / Codex** — invokes `codex exec --json -` with the prompt piped on
+  stdin, and walks the newline-delimited JSON event stream, pulling text from
+  `item.completed` and tokens from `turn.completed`. Session id is preserved.
+  Unlike the Anthropic client, there is **no separate system-prompt channel**:
+  `codex exec` has no `--system` flag, so a system prompt is prepended to the
+  user prompt in the same payload.
 
   ```yaml
   members:
@@ -244,12 +246,23 @@ Three vendor CLIs are wired:
 - **Google / Gemini** — invokes `gemini --output-format json` with
   the prompt piped on stdin, parses the `response` + `stats.models.<m>.tokens`
   envelope. OAuth consent must be granted once interactively before
-  the CLI is usable from a non-interactive shell.
+  the CLI is usable from a non-interactive shell. Like Codex and unlike Claude,
+  it has **no system-prompt flag** — `--system` is rejected outright with
+  `Unknown argument: system` — so the system prompt is prepended to the piped
+  payload.
+
+  Two caveats before enabling it, both measured 2026-08-12 and neither resolved.
+  The CLI's own `--help` says `-p/--prompt` is what selects non-interactive
+  mode, and this invocation passes neither `-p` nor the `query` positional;
+  whether piped stdin alone keeps it headless is **unestablished**. And the
+  free tier now refuses this client outright with `IneligibleTierError`, so a
+  round-trip could not be completed here at all. Enable it only if you can
+  confirm both on your own account.
 
   ```yaml
   members:
     gemini:
-      enabled: true
+      enabled: false   # see the two caveats above before flipping this
       mode: cli
       model: gemini-2.5-pro
   ```
@@ -265,6 +278,15 @@ substitutes.
 their community CLIs DO consume the API key and DO NOT bypass
 per-token billing — the contract doc warns explicitly.
 
+Both take a single `-p <prompt>` and have no second channel, so the system
+prompt is folded into that same value. Until 2026-08-12 it was **discarded**,
+and these two members answered every question with no role, no neutrality
+framing and no output contract while the run counted the reply as a peer
+verdict. Two consequences worth knowing: their whole prompt travels on argv, so
+a very large one can fail with `os_error: E2BIG` (the response carries a `hint`
+saying so), and because their CLIs report no usage block, the input side of
+their cost is an estimate rather than a provider figure.
+
 ### Cost-gate bypass for non-billable members
 
 `ExternalAIClient.billable` is the contract. Clients with

package/dist/agent-src/skills/check-refs/SKILL.md

@@ -11,6 +11,11 @@ execution:
   command:
     - ./scripts-run
     - src/scripts/check_references
+runtime_requires:
+  bins:
+    - bash
+    - node
+  network: []
 workspaces:
   - agent-config-maintainer
 packs:

package/dist/agent-src/skills/condense-memory/SKILL.md

@@ -118,7 +118,7 @@ A future `telegraph-v2.md` will tabulate the realised input-token saving against
 
 ## Cross-references
 
-- [`telegraph-speak`](../../rules/telegraph-speak.md) — runtime rule the script mirrors for input-side targets; `telegraph.speak_scope` does **not** gate this script (input-side runs regardless).
+- [`telegraph-speak`](../../rules/telegraph-speak.md) — runtime rule the script mirrors for input-side targets; that output-side rule does **not** gate this script (input-side runs regardless).
 - [`src/scripts/validate_safe_paths.ts`](../../../src/scripts/validate_safe_paths.ts) — Phase 0 gate; ported from upstream Telegraph `63a91ec`.
 - [`src/scripts/condense_memory.ts`](../../../src/scripts/condense_memory.ts) — implementation.
 - [`tests/scripts/condense_memory.test.ts`](../../../tests/scripts/condense_memory.test.ts) — regression locks for each carve-out + idempotency + denylist.

package/dist/agent-src/skills/lint-skills/SKILL.md

@@ -12,6 +12,11 @@ execution:
     - ./scripts-run
     - src/scripts/skill_linter
     - "--all"
+runtime_requires:
+  bins:
+    - bash
+    - node
+  network: []
 workspaces:
   - agent-config-maintainer
 packs:

package/dist/agent-src/skills/md-language-check/SKILL.md

@@ -11,6 +11,11 @@ execution:
   command:
     - ./scripts-run
     - src/scripts/check_md_language
+runtime_requires:
+  bins:
+    - bash
+    - node
+  network: []
 workspaces:
   - agent-config-maintainer
 packs:

package/dist/agent-src/skills/roadmap-writing/SKILL.md

@@ -22,7 +22,7 @@ packs:
 * Rewriting an existing roadmap (phase restructure, goal pivot,
   council-pass integration — not a checkbox flip)
 * Drafting a phase block, exit criteria, or rollback section that
-  will land inside an existing roadmap
+  will land inside an existing roadmap — an operator **symptom** as the input has its own entry procedure: [`symptom-driven harvest`](../../../docs/guidelines/agent-infra/symptom-driven-harvest-loop.md)
 
 Do NOT use this skill when:
 

package/dist/agent-src/skills/token-optimizer/SKILL.md

@@ -56,6 +56,10 @@ INTENT
 │     → cite markitdown (upstream: https://github.com/microsoft/markitdown)
 │       Convert FIRST, then read the markdown — never paste raw binary.
 │
+├── Large file read (source / data / log already in the tree, > 800 lines)
+│     → cite [token-efficiency § Size-gated reads](../../rules/token-efficiency.md)
+│       (size → structural grep → bounded slice; never a full read on spec)
+│
 ├── Repeated tool-call across N targets
 │     → cite [token-efficiency](../../rules/token-efficiency.md)
 │       (batch, parallelize, prune; one tool call > many)
@@ -81,7 +85,7 @@ INTENT
 |---|---|---|---|
 | `cli-output-handling` | `src/rules/cli-output-handling.md` | `verbose`, `tail`, `grep`, `CLI` | Wrap-tail-grep contract for any verbose command |
 | `rtk-output-filtering` | `src/skills/rtk-output-filtering/SKILL.md` | `rtk`, `verbose`, `filter`, `wrap` | Project-local rtk filters; wrapper command |
-| `token-efficiency` | `src/rules/token-efficiency.md` | `redirect`, `verbose`, `concise`, `tool` | Batch + parallelize tool calls; brevity floor |
+| `token-efficiency` | `src/rules/token-efficiency.md` | `redirect`, `verbose`, `concise`, `tool`, `large file` | Batch + parallelize tool calls; brevity floor; size-gated reads (probe → grep → slice above 800 lines) |
 | `agent-handoff` | `src/domains/meta/agent-handoff/command.md` | `handoff`, `fresh`, `chat`, `context`, `resume`, `session` | Session picker + generated handoff auto-seeding the next session |
 | `direct-answers` | `src/rules/direct-answers.md` | `brevity`, `flattery`, `severity`, `tiered` | Iron-Law brevity floor (kernel) |
 | `markitdown` | upstream: github.com/microsoft/markitdown | `PDF`, `DOCX`, `HTML`, `convert` | Document → markdown converter (authoritative-link only) |