neo-agent-skills 0.1.65 → 0.1.67

package/.agents/skills/epic-create/references/epic-create-workflow.md

@@ -48,7 +48,7 @@ Each sub the decomposition creates MUST be a **leaf that a single PR can FULLY d
 
 ## Procedure
 
-1. **Confirm Epic-shape, then run §1a arm `(v)`.** The work needs ≥2 coordinated subs; a single bounded artifact (≈1 PR's worth) is a standalone ticket (`ticket-create`), not an Epic. Arm `(v)` — the epic-layer sweep, mandatory before authoring — compares **terminal predicates**, not titles, reading every open `label:epic` issue's `Terminal predicate:` line and only the bodies an outcome-overlap query ranks.
+1. **Confirm Epic-shape, then run §1a arm `(v)` ([`decision-substrate-sweeps.md`](../../ticket-create/references/decision-substrate-sweeps.md#v-epic-layer-outcome-authority-sweep)).** The work needs ≥2 coordinated subs; a single bounded artifact (≈1 PR's worth) is a standalone ticket (`ticket-create`), not an Epic.
    - **Size:** Around 25 subs is the working size; an epic is a checkable outcome, and past that count the leaves no longer share one (GitHub's native sub-issue cap of 100 per parent on `POST /issues/{n}/sub_issues` is the hard wall, not the target). Count the parent's subs before filing under it (`gh api repos/<r>/issues/N/sub_issues --paginate --jq length`); past ~25, file a successor epic with its own `Terminal predicate:` and link the new leaf there. A saturated epic gets the closeout treatment: cluster its open subs into outcomes, file the successors, re-parent, and resolve the old one via `epic-resolution`.
 2. **Run the Agent OS structure map.** Before authoring the body, run `npm run ai:structure-map -- --files --loc`; use it for Agent OS / architecture placement claims or record N/A. Brain-hosted, so absent elsewhere — never add `--silent`, which hides npm's `Missing script` and leaves an empty exit-1 that reads as a pass.
 3. **Graduation gate (if from a Discussion).** High-blast Epics require the §6.2 family-keyed quorum + the §5.1 divergence matrix in the source Discussion before filing (per `ideation-sandbox-workflow.md` + `ideation-sandbox/audits/double-diamond-divergence-guard.md`). Carry the `Signal Ledger` / dissent / liveness / criteria-mapping sections into the body.

package/.agents/skills/pr-review/references/reviewer-instrument-audit.md

@@ -24,7 +24,7 @@ A qualifying claim carries: (1) **Producer** (command, MCP tool, or query class
 
 ### 3. Worked negative examples (do NOT fire)
 
-- **Routine workflow attestations:** `"Live latest-open sweep: checked latest 20 open issues at 2026-08-22T00:35Z; no equivalent found"` — a fixed gate receipt, not a population premise.
+- **Routine workflow attestations:** `"Live latest-open + closed sweep: checked latest 20 open + closed search (<problem nouns>, Out of Scope / Avoided Traps read on hits) at 2026-08-22T00:35Z; no equivalent or prior exclusion found"` — a fixed gate receipt, not a population premise.
 - **CI-linked test totals:** `"26/26 green"` or `"4,115 passed"` backed by PR CI or a linked test run — producer, tree, and scope already live in the CI receipt.
 - **Non-argument numbers:** dates, timestamps, versions, issue/PR IDs, and enumerations whose producer is already named in the surrounding table or heading.
 

package/.agents/skills/ticket-create/references/decision-substrate-sweeps.md

@@ -34,30 +34,31 @@ Fires only on **epic-labeled** filings.
 
 ```bash
 gh issue list --repo <r> --label epic --state open --json number,title
+gh issue list --repo <r> --label epic --state closed --search "<outcome nouns>" --limit 15
 ```
 
 Compare **terminal predicates**, not titles: *what state of the world does this Epic declare finished?* Two Epics that finish the same sentence are one Epic, however differently they open.
 
-**Read every predicate, then selected bodies.** `epic-create` puts `Terminal predicate:` first to keep this exhaustive read affordable:
+**Read every open predicate, then closed hits and ranked bodies.** `epic-create` puts `Terminal predicate:` first to keep the open read affordable:
 
-1. **Every row:** read each open epic's terminal predicate.
+1. **Every open row + closed search hits:** read each open epic's `Terminal predicate:` line, plus the predicate or body opening of same-outcome hits from the closed `--search`.
 2. **Ranked bodies:** for missing or arguably overlapping predicates, run the semantic outcome-overlap query and read its ranked candidates. Retrieval proposes; the reader judges overlap.
 
-**Cost control:** read predicates exhaustively, bodies selectively. On 2026-09-02, 29 Engine epics implied ~290k characters for an all-body read; re-query populations rather than calling them small. Standing outcomes can predate `(i)`'s window and use different nouns, splitting ownership when missed.
+**Cost control:** read open predicates exhaustively, closed epics and bodies selectively. On 2026-09-02, 29 Engine epics implied ~290k characters for an all-body read; re-query populations rather than calling them small. Standing outcomes can predate `(i)`'s window and use different nouns, splitting ownership when missed.
 
 > **Anchor — `neomjs/neo-agent-brain#38` vs open `#54`, 2026-08-13.** Two roots, 17 and 16 children, the same two terminal predicates. The author's §1a sweep ran honestly and **passed**; a peer's epic review caught it after filing.
 
 ## Attestation
 
-Record what ran, in the body, beside the live-open attestation:
+Record what ran, in the body, beside the live-open + closed attestation:
 
 ```
 MC sweep: <queries>, <n> results, no prior decision found
 Own-assignment sweep: <n> open, none overlapping
-Epic sweep: <n> open epics read, no shared terminal predicate    # epic-labeled filings only
+Epic sweep: <n> open predicates + closed search (<outcome nouns>) read, no shared terminal predicate    # epic-labeled filings only
 ```
 
-Prose discipline that leaves no artifact cannot be checked, and these arms exist because prose discipline failed.
+Prose discipline that leaves no artifact cannot be checked; these arms exist because prose discipline failed.
 
 ## Same-turn reconciliation
 
@@ -72,12 +73,9 @@ Reuse adjacency results and predecessor PR links: read affected tickets, open PR
 
 ## Empirical anchor
 
-**`neomjs/neo#17997`, 2026-08-31.** Re-filed **row 3** of `#17868` — open, older than the `(i)` window, and assigned to the filer — four hours after that same filer's own `#17961` closing comment recorded that row 3's blocker belonged **on `#17868`**.
-
-The missed decision can live in an older parent, a closed decline, a prior self-assignment or someone else's standing outcome; a latest-open artifact scan alone covers none of those cases.
-
-**Why the sweeps run LAST.** `#12856`, 2026-06-10: four agents raced one prompt into three duplicates, each having swept honestly; the one who re-checked immediately before filing produced none. Check-at-start decays across a multi-minute protocol.
+- **`neomjs/neo#17997`, 2026-08-31:** Re-filed **row 3** of `#17868` — open, older than `(i)`'s window, and assigned to the filer — four hours after that filer's own `#17961` closing comment recorded row 3's blocker belonged **on `#17868`**.
+- **Why sweeps run LAST (`#12856`, 2026-06-10):** four agents raced one prompt into three duplicates; the one who re-checked immediately before `create_issue` produced none.
 
 ## Sunset
 
-When `ticket-create` gains a mechanical pre-flight that runs the sweeps itself, arms `(i)`–`(iv)` collapse into that runner and this payload retires with them. `(v)` survives it as a prompt rather than a check: a runner can list open Epics, but *"do these two finish the same sentence?"* is not a judgement a lint makes.
+When `ticket-create` gains a mechanical pre-flight runner, arms `(i)`–`(iv)` collapse into it and this payload retires. `(v)` survives as a prompt: a runner can list open Epics and closed `--search` hits, but *"do these two finish the same sentence?"* is a reader's judgment.

package/.agents/skills/ticket-create/references/ticket-create-workflow.md

@@ -15,35 +15,36 @@ A ticket that calls an existing behavior wrong carries its `Design authority:` l
 **Before drafting any title or body**, you MUST execute the applicable sweeps below to ensure swarm synchronicity and architectural discipline:
 
 ### 1a. The Content Sweep (Duplicate Detection)
-Verify no equivalent ticket already exists. Redundant tickets pollute the Knowledge Base.
+Verify no equivalent ticket or closed exclusion already exists. Redundant tickets pollute the Knowledge Base.
 
 Duplicates hide in **two substrates**; sweep **both as the LAST step immediately before `create_issue`** — never at turn-start or before drafting. A pre-draft sweep goes stale while you write the Fat Ticket: in a thundering herd, a peer's ticket can land between your sweep and your `create_issue` call. The turn-start mailbox check (`§mailbox_check_protocol`) is for general coordination, **not** the dup gate.
 
-**(i) GitHub live freshness sweep (mandatory) — catches already-FILED duplicates:** read at least the latest 20 open GitHub issues from the live tracker, including issue number, title, author, labels, and URL.
+**(i) GitHub live freshness + closed-state sweep (mandatory) — catches already-FILED duplicates and closed rulings:** read at least the latest 20 open GitHub issues from the live tracker (issue number, title, author, labels, URL) and query closed state on the problem's nouns:
 
 ```bash
 gh issue list --state open --limit 20 --json number,title,author,labels,url
+gh issue list --state closed --search "<problem nouns>" --limit 15
 ```
 
-An equivalent MCP call is acceptable only when its page is live AND created-descending (`list_issues` with `sort: 'created'`); updated-descending is only supplemental — it buries untouched fresh filings below recently-updated older issues. Failure mode `#15603`: a stale filtered-read path lagged the tracker by days across this herd window. This live sweep is required even when KB and local searches return no duplicates: the most likely active-swarm duplicate can exist on GitHub before the Knowledge Base has ingested it.
+Read the `Out of Scope` and `Avoided Traps` sections of any same-surface closed hit — a closed ticket's title rarely reveals that it governs adjacent scope, and a closed hit answers what was already decided, never whether the premise matters (`neomjs/neo#18308` zero-backwards-compatibility ruling while `#18537` still closed `not_planned`, `neomjs/neo#18399` `Out of Scope` exclusion, `neomjs/neo#16391` / PR `#16403` already-shipped `hasLandedInPlace`). An equivalent MCP call for the open page is acceptable only when live AND created-descending (`list_issues` with `sort: 'created'`); updated-descending is supplemental because it buries untouched fresh filings (`#15603`). Both runs are required even when KB and local searches return no hits.
 
-If the live latest-open sweep fails because of sandbox, network, or auth state, retry through the appropriate approved/escalated path. If live GitHub state still cannot be fetched, stop before `create_issue` and report the blocker; do not file from stale-only evidence.
+If live GitHub state cannot be fetched due to sandbox, network, or auth state, retry through the approved path; if still unreachable, stop before `create_issue` and report the blocker rather than filing from stale evidence.
 
-Record the live-open sweep result in the ticket body or creation notes, e.g. `Live latest-open sweep: checked latest 20 open issues at <timestamp>; no equivalent found` or link the existing ticket you found instead of filing a duplicate.
+Record both on the same attestation line in the ticket body, e.g. `Live latest-open + closed sweep: checked latest 20 open + closed search (<problem nouns>, Out of Scope / Avoided Traps read on hits) at <timestamp>; no equivalent or prior exclusion found` (or link the existing ticket instead of filing).
 
-**(ii) A2A in-flight claim sweep (mandatory) — catches IN-FLIGHT duplicates not yet on GitHub:** for the first minutes of a thundering herd a peer's intent exists only as an A2A `[lane-claim]`/`[lane-intent]`, not yet a GitHub ticket — the `(i)` sweep above cannot see it. Immediately before `create_issue`, scan the mailbox for recent claims on the same scope:
+**(ii) A2A in-flight claim sweep (mandatory) — catches IN-FLIGHT duplicates not yet on GitHub:** in a thundering herd a peer's intent first exists only as an A2A `[lane-claim]`/`[lane-intent]` invisible to `(i)`. Immediately before `create_issue`, scan the mailbox for recent claims on the same scope:
 
 ```js
 list_messages({ status: 'all', limit: 30 })  // ALL read-states — recency is the bound, not read-status
 ```
 
-**Filter by recency + scope, NOT read-status.** A `[lane-claim]` you have already *read* (e.g. your turn-start mailbox triage marked it read) is still an active claim — `status: 'unread'` would make it invisible and you would refile the duplicate, the exact late-boot ordering this gate exists to stop (read-status is consumption tracking, not a relevance signal). From the returned slice, keep messages whose `sentAt` falls in the herd window (~last 30–60 min) **and** whose `[lane-claim]`/`[lane-intent]` subject overlaps your scope (`taggedConcepts: ['lane-claim']` is optional narrowing — tag discipline is conventional, not guaranteed, so subject-scan stays primary).
+**Filter by recency + scope, NOT read-status.** A `[lane-claim]` already marked read during turn-start mailbox triage is still an active claim — `status: 'unread'` hides it and causes duplicate filing (read-status tracks consumption, not relevance). Keep messages whose `sentAt` falls in the herd window (~last 30–60 min) **and** whose `[lane-claim]`/`[lane-intent]` subject overlaps your scope (`taggedConcepts: ['lane-claim']` is optional; subject-scan stays primary).
 
-**Tiebreak — first-claim-timestamp-wins:** if a competing claim or a just-filed ticket surfaces, the **earliest claim/file timestamp wins** — and an earlier *unfiled claim* outranks a later *filed ticket* (claims must bind against fast-filers, or the sweep is toothless). The later party stands down; if it already filed, it closes its duplicate and ports any unique substance onto the survivor — i.e. the earlier claimant absorbs the later filer's content, **not** the reverse. Deterministic and self-healing — resolves simultaneous filings without lead mediation. A contested-lane resolution MUST set `wakeSuppressed: false` — claim-class is quiet by default (#15987), so omission leaves the "do-not-re-file" signal mailbox-only and it reaches no one in time.
+**Tiebreak — first-claim-timestamp-wins:** the **earliest claim/file timestamp wins**, and an earlier *unfiled claim* outranks a later *filed ticket*. The later party stands down; if it already filed, it closes its duplicate and ports unique substance onto the survivor (earlier claimant absorbs later filer's content, **not** the reverse). A contested-lane resolution MUST set `wakeSuppressed: false` — claim-class is quiet by default (#15987), so omitting it leaves the signal mailbox-only.
 
-**(iii) Memory Core rationale sweep + (iv) own-assignment sweep (both mandatory):** `(i)`, `(ii)` and the greps below are all **artifact** substrates — they answer *"does a ticket exist?"*, never *"was this already decided, and why?"* Run one `query_raw_memories` keyed on the **problem's** nouns (the symptom as observed, never the mechanism you are about to build), and `gh issue list --state open --assignee @me` reading the **bodies** of same-surface hits. Both attest in the body like `(i)` does.
+**(iii) Memory Core rationale sweep + (iv) own-assignment sweep (both mandatory):** `(i)` and `(ii)` are **artifact** substrates — combine them with one `query_raw_memories` keyed on the **problem's** nouns (observed symptom, never proposed mechanism) and `gh issue list --state open --assignee @me` reading the **bodies** of same-surface hits. Both attest in the body beside `(i)`.
 
-**(v) Epic-layer sweep (epic-labeled filings only, mandatory):** both artifact axes are blind to an outcome authority — the payload says why. Read every open `label:epic` issue's `Terminal predicate:` line, then body-read only what an outcome-overlap query ranks, and attest beside `(i)`.
+**(v) Epic-layer sweep (epic-labeled filings only, mandatory):** read every open `label:epic` issue's `Terminal predicate:` line, run `gh issue list --label epic --state closed --search "<outcome nouns>"` to read the predicate or body opening of same-outcome closed epics, body-read what an outcome-overlap query ranks, and attest beside `(i)`.
 
 <!-- trigger: filing a ticket, filing an EPIC, or a defect found while measuring something else -> read ./decision-substrate-sweeps.md (query shape, the epic sweep, attestation lines, same-turn porting, the #17997 / #12856 anchors) -->
 <!-- trigger: load-bearing population count or absence claim justifying premise, scope, or acceptance -> read ../../pr-review/references/reviewer-instrument-audit.md (role-triggered publish-safe instrument receipt; routine receipts and CI totals exempt) -->

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "neo-agent-skills",
-  "version": "0.1.65",
+  "version": "0.1.67",
   "description": "Install the workflow substrate a cross-model AI team actually runs on: ticket intake, PR review gates, lane coordination and guards, versioned as a package. The skills Neo.mjs maintainers load every session — materialized into your repo, not copied.",
   "type": "module",
   "license": "MIT",