task-pipeline-skill 1.102.3 → 1.102.4

package/CHANGELOG.md

@@ -1,3 +1,26 @@
+## v1.102.4: the Bash hooks start no process in a project with no run
+
+- **Fewer processes per Bash call, same verdicts** (hub board row B-310). The plugin's Bash
+  hooks run on every Bash call in every project, and they started processes before learning
+  there was no run. Measured by PID delta (macOS allocates PIDs in order; minimum of 25 runs,
+  the hook's own bash included): `release-gate.sh` 14, `gate-observer.sh` 3 and
+  `memory-nudge.sh` 1, so 18 per Bash call in a project with no `.task-pipeline/run.md`, and 23
+  with a run in flight (14, 4 and 5). Now 3 (1, 1, 1) and 11 (5, 4, 2). Three changes:
+  `release-gate.sh` tests for the ledger before it reads stdin or starts python3 (`-e`, as
+  the python asks `os.path.exists`, so a ledger path that is a directory is still refused as
+  unreadable); it splits its `state`, `act`, `why` verdict with parameter expansion instead
+  of three `printf | cut` pipelines, keeping `cut`'s field rules, and a verdict of more than
+  one line still goes through the pipelines; `gate-observer.sh` reads stdin after the ledger
+  test; `memory-nudge.sh` finds its scripts directory from `${BASH_SOURCE[0]}` without
+  `$(cd "$(dirname …)" && pwd)`. Behaviour is unchanged: a differential run of the old and
+  new hooks over 4,950 cases (eight ledger states, five flows, 30 payloads, python3 working
+  and broken) gave 0 differences in exit code, stdout, stderr and ledger contents.
+  `test/release_gate_test.py` (74, three new): with no ledger each Bash hook exits 0 silent and starts
+  none of python3, cat, cut, dirname (PATH shims log any start; red on 1.102.3, which started
+  cat and python3), a negative control that the shims do see python3, and a refusal that
+  carries its act and whole reason. `test/memory_nudge_test.py` (43, two new): the run marker is still
+  honoured when the hook is started by a bare name or a relative path.
+
 ## v1.102.3: stamp rows keep their table's shape, one rotation line per quarter, apply_patch gated
 
 Guards: 531 → **531**; property checks 15 → **15**. No negative was added: the new checks are

package/SKILL-CARD.md

@@ -12,7 +12,7 @@ harmless.
 |---|---|
 | **Purpose** | Runs a substantial task through ten gated delivery stages — intake grill, docs study, brainstorm, spec, plan, subagent build, tests, lint/deploy, post-deploy, docs+registers, acceptance — refusing to advance until each gate passes |
 | **Owner** | ssheleg ([github.com/ssheleg/task-pipeline](https://github.com/ssheleg/task-pipeline)) |
-| **Version** | 1.102.3 |
+| **Version** | 1.102.4 |
 | **Surface** | Claude Code (filesystem skill + plugin) and the vercel `skills` CLI. **Not** uploaded to the Skills API; custom Skills do not sync across surfaces |
 | **Dependencies** | None required. Optional: `context7` (MCP), `figma` (MCP), super-ux, agent-sync, graphify, obsidian-wiki, and **one of two browser channels** — `playwright` (CLI or MCP) or `chrome-devtools` (MCP); either satisfies the browser step and neither is required. The engine is WebKit (Safari) first; `chrome-devtools` is Chrome-only, a stated exception. Every stage's doctrine ships in-repo; user-facing stage 3 requires a valid scenario artifact; super-ux is preferred, with equivalent-provider and inline-fallback paths under the provider contract |
 | **Evaluation status** | Suite authored, 5 categories. One recorded run, **self-observed by the author**; **zero blind runs on zero of three models** — the split, and the numbers, live in [`evals/RESULTS.md`](evals/RESULTS.md) and are computed by `evals/run.py` |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "task-pipeline-skill",
-  "version": "1.102.3",
+  "version": "1.102.4",
   "description": "Full-cycle delivery pipeline for coding agents: a mandatory built-in intake grill, then 10 gated stages (docs, brainstorm+decompose, spec, plan, build, tests, lint/deploy, post-deploy, docs/wiki, acceptance). Every stage's doctrine ships inside the skill — no companion plugin required. This package is the installer CLI.",
   "bin": {
     "task-pipeline": "bin/task-pipeline.js"

package/plugins/task-pipeline/.claude-plugin/plugin.json

@@ -3,7 +3,7 @@
   "name": "task-pipeline",
   "displayName": "Task Pipeline",
   "description": "Runs a substantial task through a mandatory built-in intake grill, then 10 gated stages (docs, brainstorm+decompose, spec, plan, subagent build, tests, lint/deploy, post-deploy, docs/wiki, acceptance). Every stage's doctrine is built into the skill — no companion plugin required — with typed auto/judgment/manual gates, a frozen requirement spine that closes with evidence, a work board and a verification ledger that outlive a run, an exposure line naming what shipped unconfirmed, a progress rail computed from the project's own config, a proportional short path for small changes, a loop guard whose ceilings stop and record a blocker, and stage-3 tracks for what a product does, how it sounds and how it looks. Two modes need no task: `checkup` (what is unverified) and `setup` (audit existing docs); a third, `waves`, takes the whole board and closes it in waves that each finish in one pass. Retro insights can publish upstream as issues, opt-in and redacted.",
-  "version": "1.102.3",
+  "version": "1.102.4",
   "author": {
     "name": "ssheleg",
     "url": "https://x.com/sshlg93"

package/plugins/task-pipeline/hooks/gate-observer.sh

@@ -26,10 +26,12 @@
 # running a pipeline.
 set -uo pipefail
 
-input=$(cat 2>/dev/null || true)
 project="${CLAUDE_PROJECT_DIR:-$PWD}"
 ledger="$project/.task-pipeline/run.md"
+# The ledger test comes before stdin is read (hub B-310): with no run, reading the
+# payload starts `cat` for nothing. memory-nudge.sh has always exited unread here.
 [ -f "$ledger" ] || exit 0
+input=$(cat 2>/dev/null || true)
 
 HOOK_INPUT="$input" python3 - "$ledger" "$project" <<'PY' 2>/dev/null || true
 import json, os, re, sys, datetime

package/plugins/task-pipeline/hooks/memory-nudge.sh

@@ -20,7 +20,14 @@
 set -uo pipefail
 ledger="${CLAUDE_PROJECT_DIR:-$PWD}/.task-pipeline/run.md"
 [ -f "$ledger" ] || exit 0
-TP_SCRIPTS="$(cd "$(dirname "${BASH_SOURCE[0]}")/../skills/task-pipeline/scripts" 2>/dev/null && pwd)" \
+# The scripts directory without `$(cd "$(dirname …)" && pwd)`, which started two
+# processes per call (hub B-310). `dirname` of a bare name is `.`; a directory that
+# does not exist leaves TP_SCRIPTS empty, as the failed `cd` did.
+here=${BASH_SOURCE[0]}
+case "$here" in */*) here=${here%/*} ;; *) here=. ;; esac
+TP_SCRIPTS="$here/../skills/task-pipeline/scripts"
+[ -d "$TP_SCRIPTS" ] || TP_SCRIPTS=
+TP_SCRIPTS="$TP_SCRIPTS" \
 python3 - "$ledger" <<'PY' 2>/dev/null || true
 import hashlib, json, os, re, sys
 ledger = sys.argv[1]

package/plugins/task-pipeline/hooks/release-gate.sh

@@ -43,11 +43,19 @@
 # crashing gate that fails open is worse than no gate, because it reads as one.
 set -uo pipefail
 
-input=$(cat 2>/dev/null || true)
-
 project="${CLAUDE_PROJECT_DIR:-$PWD}"
 ledger="$project/.task-pipeline/run.md"
 
+# Narrowness 2, before anything else starts (hub B-310): with no ledger every path
+# below ends in `skip` or `exit 0` with no output, so the answer is known without
+# reading stdin or starting python3. `-e`, not `-f`: the python below asks
+# `os.path.exists`, so a ledger path that is a directory still reaches it and is
+# still refused as unreadable. Leaving stdin unread is what memory-nudge.sh has
+# always done on a project with no run.
+[ -e "$ledger" ] || exit 0
+
+input=$(cat 2>/dev/null || true)
+
 decide() {
   # The payload travels in the environment, NOT on stdin: the heredoc below IS
   # stdin for `python3 -`, so a script fed that way can never also read the hook's
@@ -265,9 +273,27 @@ if [ $status -ne 0 ] || [ -z "$verdict" ]; then
   exit 2
 fi
 
-state=$(printf '%s' "$verdict" | cut -f1)
-act=$(printf '%s' "$verdict" | cut -f2)
-why=$(printf '%s' "$verdict" | cut -f3)
+# The verdict is `state<TAB>act<TAB>why`, split in the shell itself rather than by
+# three `printf | cut` pipelines (nine processes per call, hub B-310). Each value is
+# what `cut -fN` gives on one line: a line with no tab is its own field 1, 2 and 3;
+# a line with one tab has an empty field 3; a field ends at the next tab. A verdict
+# of more than one line (a newline inside a declared command or stage id) keeps the
+# original pipelines, so that case reads exactly as before.
+case "$verdict" in
+  *$'\n'*)
+    state=$(printf '%s' "$verdict" | cut -f1)
+    act=$(printf '%s' "$verdict" | cut -f2)
+    why=$(printf '%s' "$verdict" | cut -f3) ;;
+  *$'\t'*)
+    state=${verdict%%$'\t'*}
+    rest=${verdict#*$'\t'}
+    act=${rest%%$'\t'*}
+    case "$rest" in
+      *$'\t'*) rest=${rest#*$'\t'}; why=${rest%%$'\t'*} ;;
+      *) why= ;;
+    esac ;;
+  *) state=$verdict act=$verdict why=$verdict ;;
+esac
 
 case "$state" in
   skip|ok) exit 0 ;;

package/plugins/task-pipeline/skills/evidence-docs/references/hooks.md

@@ -254,7 +254,7 @@ reusable part:
 | Narrowness | Why it is not a smaller feature |
 |---|---|
 | Only outward acts, never ordinary commits | stage 5 commits per task by design; a gate that fights the build loop is removed within a day |
-| Silent where no `.task-pipeline/run.md` exists | enabling the plugin must change nothing in a repository that runs no pipeline |
+| Silent where no `.task-pipeline/run.md` exists, before python3 starts | enabling the plugin must change nothing in a repository that runs no pipeline, and every Bash call there pays these hooks: each Bash hook tests for the ledger before it reads stdin or starts python3, so one Bash call costs three bash processes instead of eighteen (hub B-310; `test/release_gate_test.py`, "with no ledger, the Bash hooks exit 0 silent and start no process") |
 | Reads the ledger, never reruns a suite | `progress.md` already makes the ledger append-only; a second source of truth about "did stage 6 pass" is the failure this file warns about below |
 
 **Fail-closed, deliberately.** Every non-zero exit code other than `2` is

package/plugins/task-pipeline/skills/task-pipeline/references/hooks.md

@@ -254,7 +254,7 @@ reusable part:
 | Narrowness | Why it is not a smaller feature |
 |---|---|
 | Only outward acts, never ordinary commits | stage 5 commits per task by design; a gate that fights the build loop is removed within a day |
-| Silent where no `.task-pipeline/run.md` exists | enabling the plugin must change nothing in a repository that runs no pipeline |
+| Silent where no `.task-pipeline/run.md` exists, before python3 starts | enabling the plugin must change nothing in a repository that runs no pipeline, and every Bash call there pays these hooks: each Bash hook tests for the ledger before it reads stdin or starts python3, so one Bash call costs three bash processes instead of eighteen (hub B-310; `test/release_gate_test.py`, "with no ledger, the Bash hooks exit 0 silent and start no process") |
 | Reads the ledger, never reruns a suite | `progress.md` already makes the ledger append-only; a second source of truth about "did stage 6 pass" is the failure this file warns about below |
 
 **Fail-closed, deliberately.** Every non-zero exit code other than `2` is