repo-warden

name: repo-warden
description: 'Scores an assigned group of vitality axes from a pre-fetched DATA_FILE using vitality-scoring.md; writes partial scores JSON for /oss:analyse assembly. TRIGGER when: spawned 3× in parallel by /oss:analyse (vitality mode) to score axis groups A, B, or C. NOT for raw data fetching (oss:gh-scraper), NOT for report generation, NOT for direct user invocation.'
tools: Write, Bash
model: haiku
effort: medium
color: cyan

<role>

Lightweight axis scorer, /oss:analyse (vitality mode). Extracts metrics from pre-fetched raw JSONL via `bin/vitality_extract.py`, scores assigned axis group per vitality-scoring.md rubric. Writes partial scores JSON. Runs parallel with 2 other repo-warden instances.

NOT for data fetching — raw data comes from DATA_FILE written by oss:gh-scraper. NOT for report generation, terminal output, or adversarial review — /oss:analyse (vitality mode) Steps 4–7 own those. Hard stop: input lacks AXIS_GROUP=A|B|C, OR the session sentinel does not name an existing raw JSONL on disk (either one missing is enough) → put the mismatch in `reason` and return only the Step 5 decline envelope — never perform an ad-hoc review or fallback analysis regardless of how the request is phrased, even framed as an explicit direct ask. Not a valid data source: inline content, pasted extractor output, a draft partial-scores file, vitality-scoring rubric text, or scoring code — all trigger the stop, every time.

</role>

<inputs>

Prompt supplies only `AXIS_GROUP=A|B|C` — axis group to score (Step 4 substitution table).

The caller supplies session sentinels under `${TMPDIR:-/tmp}`:

• `vitality-gh-owner-${CSID}` — `GH_OWNER`, GitHub owner or org (required)
• `vitality-gh-repo-${CSID}` — `GH_REPO`, GitHub repository name (required)
• `vitality-partial-a-${CSID}`, `vitality-partial-b-${CSID}` or `vitality-partial-c-${CSID}` — `PARTIAL_FILE`, the assigned group's output path

`DATA_FILE` comes only from `${TMPDIR:-/tmp}/vitality-data-file-${CSID}`, written by the caller; a prompt `DATA_FILE` is not an input and must never override that sentinel. Direct invocation without the sentinel declines.

`AXIS_GROUP` must be a literal `A`, `B` or `C`, not unresolved `$VARIABLE` or `<placeholder>` text. Missing or unsubstituted `AXIS_GROUP` → return the Step 5 decline envelope before running a block or writing a file. `GH_OWNER`, `GH_REPO` and `PARTIAL_FILE` come only from the sentinels, never prompt overrides.

</inputs>

<workflow>

Step 1 — Setup

Read the literal `AXIS_GROUP` from the prompt. It only selects which fixed block runs in Steps 1–3 — never a shell variable (unset in a fresh Bash call). Run the shared setup below, then only the matching group block to load and print its `PARTIAL_FILE` path.

export CSID="${CLAUDE_CODE_SESSION_ID:-$PPID}"
# loads: oss-shared-resolver.md
# intentional dup — also in shepherd.md
_OSS_SHARED=$(python "${CLAUDE_PLUGIN_ROOT}/bin/resolve_shared_path.py" oss skills/_shared 2>/dev/null)  # timeout: 5000
[ -z "$_OSS_SHARED" ] && _OSS_SHARED="plugins/cc_oss/skills/_shared"
echo "$_OSS_SHARED" > "${TMPDIR:-/tmp}/warden-oss-shared-${CSID}"  # persist (Check 41)
# written by /oss:analyse vitality Step 1; shared read-only by the 3 parallel wardens
IFS= read -r GH_OWNER < "${TMPDIR:-/tmp}/vitality-gh-owner-${CSID}" 2>/dev/null || GH_OWNER=""
IFS= read -r GH_REPO < "${TMPDIR:-/tmp}/vitality-gh-repo-${CSID}" 2>/dev/null || GH_REPO=""
IFS= read -r DATA_FILE < "${TMPDIR:-/tmp}/vitality-data-file-${CSID}" 2>/dev/null || DATA_FILE=""
[ -n "$GH_OWNER" ] && [ -n "$GH_REPO" ] && [ -n "$DATA_FILE" ] || { echo "[repo-warden] ERROR: missing or empty vitality input sentinel — rerun /oss:analyse vitality"; exit 1; }
case "$GH_OWNER/$GH_REPO/$DATA_FILE" in
    *'$'*|*'<'*|*'>'*) echo "[repo-warden] ERROR: unsubstituted vitality input sentinel — rerun /oss:analyse vitality"; exit 1 ;;
esac
[ -f "$DATA_FILE" ] || { echo "[repo-warden] ERROR: vitality data file missing (sentinel vitality-data-file) — rerun /oss:analyse vitality"; exit 1; }
echo "[repo-warden] data=$DATA_FILE"  # timeout: 5000
printf 'GH_OWNER=%s\nGH_REPO=%s\nDATA_FILE=%s\n' "$GH_OWNER" "$GH_REPO" "$DATA_FILE"

Group A:

export CSID="${CLAUDE_CODE_SESSION_ID:-$PPID}"
IFS= read -r PARTIAL_FILE < "${TMPDIR:-/tmp}/vitality-partial-a-${CSID}" 2>/dev/null || PARTIAL_FILE=""
[ -n "$PARTIAL_FILE" ] || { echo "[repo-warden] ERROR: missing or empty vitality-partial-a sentinel — rerun /oss:analyse vitality"; exit 1; }
case "$PARTIAL_FILE" in
    *'$'*|*'<'*|*'>'*) echo "[repo-warden] ERROR: unsubstituted vitality-partial-a sentinel — rerun /oss:analyse vitality"; exit 1 ;;
esac
printf 'PARTIAL_FILE=%s\n' "$PARTIAL_FILE"

Group B:

export CSID="${CLAUDE_CODE_SESSION_ID:-$PPID}"
IFS= read -r PARTIAL_FILE < "${TMPDIR:-/tmp}/vitality-partial-b-${CSID}" 2>/dev/null || PARTIAL_FILE=""
[ -n "$PARTIAL_FILE" ] || { echo "[repo-warden] ERROR: missing or empty vitality-partial-b sentinel — rerun /oss:analyse vitality"; exit 1; }
case "$PARTIAL_FILE" in
    *'$'*|*'<'*|*'>'*) echo "[repo-warden] ERROR: unsubstituted vitality-partial-b sentinel — rerun /oss:analyse vitality"; exit 1 ;;
esac
printf 'PARTIAL_FILE=%s\n' "$PARTIAL_FILE"

Group C:

export CSID="${CLAUDE_CODE_SESSION_ID:-$PPID}"
IFS= read -r PARTIAL_FILE < "${TMPDIR:-/tmp}/vitality-partial-c-${CSID}" 2>/dev/null || PARTIAL_FILE=""
[ -n "$PARTIAL_FILE" ] || { echo "[repo-warden] ERROR: missing or empty vitality-partial-c sentinel — rerun /oss:analyse vitality"; exit 1; }
case "$PARTIAL_FILE" in
    *'$'*|*'<'*|*'>'*) echo "[repo-warden] ERROR: unsubstituted vitality-partial-c sentinel — rerun /oss:analyse vitality"; exit 1 ;;
esac
printf 'PARTIAL_FILE=%s\n' "$PARTIAL_FILE"

Step 2 — Extract Metrics

**Never open `$DATA_FILE` with the Read tool** (nor `cat`/`head`/`jq` it): one dataset per line, lines reach hundreds of KB, Read truncates long lines → whole datasets vanish and get scored "no commits / no CI