task-pipeline-skill 1.95.0 → 1.96.0

package/plugins/task-pipeline/skills/task-pipeline/templates/brief.md

@@ -18,6 +18,7 @@
 - Documentation (the phase-1b inventory — the four questions)
 - Scope
 - Requirements (the REQ spine — every later stage traces to these IDs)
+- Finish line (when this iteration stops)
 - Users & context
 - Decisions locked (the grill's output)
 - Autonomy (the sweep — stages 1→10 read this instead of asking)
@@ -101,6 +102,19 @@ not continuously): `open` → `planned` → `built` → `verified` \| `partial`
 > task in smaller terms is the failure this table exists to prevent: every gate
 > after it goes green on the shrunken task and nothing reports the loss.
 
+## Finish line (when this iteration stops)
+
+The contract the iteration stops against (the skill's `references/finish-line.md`). Written
+here, confirmed with the brief, and revised only as a recorded revision — a finding is not one.
+
+- **Objective:** <one sentence: what is true when this iteration is done>
+- **Exit checks:** <REQ id → command, run on the artefact the operator uses: released /
+  installed / deployed — one line each>
+- **Scope:** <the modules, paths and REQ ids this iteration may change, as a list>
+- **Budget:** `maxRounds` 3 · `maxReleases` 2 (the defaults; `pipeline.json` →
+  `run.finish` overrides them)
+- **Base:** <the commit this iteration started from>
+
 ## Users & context
 
 - **Who / for what:** … (personas, the job being done)

package/plugins/task-pipeline/skills/task-pipeline/templates/inventory.schema.json

@@ -2,7 +2,7 @@
   "$schema": "https://json-schema.org/draft/2020-12/schema",
   "$id": "https://github.com/ssheleg/task-pipeline/blob/main/plugins/task-pipeline/skills/task-pipeline/templates/inventory.schema.json",
   "title": "inventory/1 — <artifacts>/specs/YYYY-MM-DD-<topic>-inventory.json",
-  "description": "Wave 0's single machine-readable inventory: one item per open item of the board, every survey and every harvested source (references/waves.md, Wave 0). scripts/waves.py merges audit outputs into it (`merge`), writes `module`, `node`, `wave` and `release` from two short tables (`assign`), counts it (`count`) and refuses an item that lands nowhere (`validate`). The document is an object carrying `schema` and `items`; a bare array of items is read as the items of one. Fields a project needs beyond these go under an `x-` prefix.",
+  "description": "Wave 0's single machine-readable inventory: one item per open item of the board, every survey and every harvested source (references/waves.md, Wave 0). scripts/waves.py merges audit outputs into it (`merge`), writes `module`, `node`, `wave` and `release` from two short tables (`assign`), counts it (`count`), refuses an item that lands nowhere (`validate`) and computes the finish-line verdict from the triaged items (`finish`, references/finish-line.md). The document is an object carrying `schema` and `items`; a bare array of items is read as the items of one. Fields a project needs beyond these go under an `x-` prefix.",
   "type": "object",
   "additionalProperties": false,
   "required": ["schema", "items"],
@@ -17,6 +17,14 @@
       "additionalProperties": false,
       "patternProperties": {"^x-[a-z0-9-]+$": {}},
       "required": ["id", "title", "area", "kind", "severity", "evidence", "proposed_fix"],
+      "allOf": [
+        {"if": {"required": ["triage"], "properties": {"triage": {"const": "blocking"}}},
+         "then": {"required": ["test", "round"]}},
+        {"if": {"required": ["test"]},
+         "then": {"required": ["triage"], "properties": {"triage": {"const": "blocking"}}}},
+        {"if": {"required": ["test", "base_repro"], "properties": {"test": {"const": "B3"}}},
+         "then": {"properties": {"base_repro": {"not": {"const": "yes"}}}}}
+      ],
       "properties": {
         "id": {
           "description": "The item's id in its owning register. An item without one gets one there, in the same change. Unique across the inventory.",
@@ -60,6 +68,12 @@
         "blocked_by": {"type": "array", "items": {"type": "string"}},
         "closable_in_one_pass": {"type": "boolean"},
         "needs_operator": {"type": "boolean"},
+        "triage": {"description": "The finish-line class, set when the finding is found (references/finish-line.md → Triage at discovery). Only blocking findings argue for another round; deferred and noise never affect `waves.py finish`.", "enum": ["blocking", "deferred", "noise"]},
+        "test": {"description": "The blocking test that matched, first match wins: B1 an exit check is red; B2 a hotfix-lane entry; B3 it passes at Base; B4 a verified REQ or a published claim is false on the artefact. Required on a blocking item, refused on any other.", "enum": ["B1", "B2", "B3", "B4"]},
+        "round": {"description": "The finish-line round that found it, from 1. Required on a blocking item.", "type": "integer", "minimum": 1},
+        "base_repro": {"description": "Does the failure reproduce at the iteration's Base? `no` is a finding this iteration caused (self-inflicted).", "enum": ["yes", "no", "n/a"]},
+        "blast": {"description": "Who it hurts, 1–3 (references/prioritisation.md, backlog `sev × blast`): a priority input of a deferred row. `waves.py finish` lists a deferred item at blast 3 as a known risk.", "type": "integer", "minimum": 1, "maximum": 3},
+        "rung": {"description": "The impact rung (references/prioritisation.md): a priority input of a deferred row. `waves.py finish` lists a deferred item at I4 or I5 as a known risk.", "enum": ["I1", "I2", "I3", "I4", "I5"]},
         "notes": {"type": "string"}
       }
     },

package/plugins/task-pipeline/skills/task-pipeline/templates/run.md

@@ -20,7 +20,7 @@ Run: `<topic>` · started `<YYYY-MM-DD>` · module map: `<path or "none">`
 
 ## `read:` — which doctrine this run actually opened
 
-The bundle is 52 reference files and nothing recorded which of them a run read, so **a
+The bundle is 53 reference files and nothing recorded which of them a run read, so **a
 skipped file and a read one were indistinguishable** — the class every guard in this
 pipeline exists to catch, left standing over the doctrine itself.
 
@@ -48,7 +48,7 @@ same claim one level down.
 |---|---|---|
 | `unmeasured — no run ledger` | there is no ledger | nothing to read from |
 | `unmeasured — the ledger carries no read: lines` | the hook is absent, **or** the run opened no doctrine | two opposite facts, and the ledger cannot separate them, so neither is claimed |
-| `N of 52 reference files read — unattested`, then each unread one | the hook is installed and fired | the count alone says there is a gap, not where — and `unattested` says the ledger cannot name who wrote the lines |
+| `N of 53 reference files read — unattested`, then each unread one | the hook is installed and fired | the count alone says there is a gap, not where — and `unattested` says the ledger cannot name who wrote the lines |
 
 **It is a disclosure: no floor, no direction, never a target.** A run that needs four files
 and reads four is not worse than one that reads thirty — and the moment the number becomes
@@ -71,6 +71,9 @@ holds: <stage id> — <n> (<class: what, owner>; … or "none") — enumerated <
 gate:  <stage id> — command "<cmd>" — exit <N> — <ISO-8601>
 event: <compact|session-end|subagent|memory> — <detail> — <ISO-8601>
 review: <stage id> — surface <name> — rounds <N> — <returned|approved|unresolved> — <ISO-8601>
+finding: <id> — blocking:B<n> | deferred:B-NNN | noise — round <N> — base-repro yes|no|n/a
+round: <N> — <reading> — new blocking <k>
+finish: done|budget|blocked — rounds <n>/<max> — releases <n>/<max>
 read:  references/<file>.md            # hook-appended, deduped, UNATTESTED (no writer field)
 ```
 
@@ -118,6 +121,20 @@ read:  references/<file>.md            # hook-appended, deduped, UNATTESTED (no
   into the checkpoint of the stage it names, and stage 10 copies the last one per surface
   into the acceptance file. **It is a measurement, never a target**: a number of passes to
   beat is an instruction to stop showing the sheet.
+- **`finding:`** — one line per finding, written when it is found and before anyone fixes
+  it (`references/finish-line.md` → *Triage at discovery*, inside the skill). The class is
+  one of `blocking`, `deferred`, `noise`; a blocking line names its test, one of `B1`, `B2`,
+  `B3`, `B4`; a deferred line names its board row. `base-repro` says whether the failure
+  reproduces at the iteration's Base — `no` is a finding this iteration caused.
+- **`round:`** — one line per finish-line round, **clean rounds included**: its number, the
+  independent reading it ran (an audit axis, an observer, the post-release readback) and how
+  many new blocking findings it produced. A clean round leaves no `finding:` line, so without
+  this line two clean rounds and no rounds look the same; `scripts/waves.py finish --ledger`
+  counts the rounds from it and refuses when its counts disagree with the inventory.
+- **`finish:`** — one line when the iteration stops, beside stage 10's verdict: `done`,
+  `budget` or `blocked`, with the rounds and releases spent against their budget. The
+  block under it is `references/finish-line.md` → *The finish block*;
+  `scripts/waves.py finish` computes the verdict from the inventory.
 - **`touch:`** — one line per file per pass, and the reason names **what forced the
   edit**: a finding id, a failed gate item, an operator instruction. *"Cleanup"*,
   *"polish"* and *"while I was there"* are not reasons; they are churn with better
@@ -141,11 +158,18 @@ event: compact — auto — 2026-08-10T11:58Z
 event: subagent — general-purpose — 2026-08-10T12:00Z
 gate:  6 — command "npm test" — exit 0 — 2026-08-10T12:02Z
 review: 6 — surface landing — rounds 1 — returned — 2026-08-10T12:02Z
+finding: F-021 — blocking:B3 — round 1 — base-repro no
+finding: F-022 — deferred:B-118 — round 1 — base-repro yes
+round: 1 — exit checks + audit axis seams — new blocking 1
+finding: F-023 — noise — round 2 — base-repro n/a
+round: 2 — exit checks + observer — new blocking 0
+round: 3 — exit checks + post-release readback — new blocking 0
 stage: 6 Tests — gate manual — verdict pass — 2026-08-10T12:03Z
 hand:  3 — task "add CSV export to the orders table" — done 2 — surfaced 1 — decisions 1 — amb 2 (OQ-0007, ledger row 4)
        scope 5f21ac3/node-24-linux/REQ-001,REQ-004 — unverified 1 (XLSX path: no fixture)
 holds: 5 — 2 (worktree: build-csv-export, this run; container: pg-test, this run) — enumerated 8/8 classes
 holds: 10 — none — enumerated 7/8 classes, unlooked: containers (no docker on this host)
+finish: done — rounds 3/3 — releases 1/2
 ```
 
 The last two lines are a **trip**: the same file, two consecutive passes, the same