git:20260821.d979319 to git:20260822.a5f4fd6

+ If `$OSS_AVAILABLE` is `true`: call `AskUserQuestion` ONCE with BOTH questions in the same call (a second sequential window costs another human-idle round trip) — Q1: "Looks like you passed a PR/issue number. Did you mean to run `/oss:review $ARGUMENTS` (requires oss plugin) to review that PR?" Options: (a) "Yes — launch `/oss:review $ARGUMENTS`" · (b) "No — review local code". Q2: "If reviewing local code: provide the file path or directory (free text; skip if Q1 = Yes)". On (a): call `Skill(skill="oss:review", args="$ARGUMENTS")`, ignore Q2. On (b): use the Q2 response as `$REVIEW_ARGS` and proceed to Step 1 (Q2 empty → ask once more for the path).

+ If `$OSS_AVAILABLE` is `false`: call `AskUserQuestion` ONCE with BOTH questions — Q1: "Looks like you passed a PR/issue number, but oss plugin not installed — `/oss:review` unavailable. Review local code instead?" Options: (a) "Yes — review local code" · (b) "I need oss plugin". Q2: "If reviewing local code: provide the file path or directory (free text; skip if Q1 = b)". On (a): use the Q2 response as `$REVIEW_ARGS` and proceed to Step 1 (Q2 empty → ask once more). On (b): inform user: install with `claude plugin install oss@borda-ai-rig`.

+ FANOUT_MAX=3 # default: top-N spawn UNITS among classification-preselected dimensions

+ # cap counts dimension spawn units only; real default lineup = up to 3

+ # dimension units + challenger + bridge co-review + consolidator

+ # (+ up to 3 cross-validation verifiers when criticals exist)

+ # --full runs ALL preselected units instead — no numeric cap

+ Scope + non-Python impact check + Python filter in ONE pass — a single `git diff` capture feeds TARGET, the report-header warnings, and the early exit (previously eight diff invocations across three blocks):

+

+ # timeout: 5000

+ _DIFF_FILES=$(git diff --name-only HEAD 2>/dev/null)

+ PYTHON_FILES=$(find "$REVIEW_ARGS" -name '*.py' -type f 2>/dev/null | head -1)

+ echo "$_DIFF_FILES"

+ TARGET="working-tree diff ($(echo "$_DIFF_FILES" | grep -c '\.py$' | tr -d ' ') Python files)"

+ PYTHON_FILES=$(echo "$_DIFF_FILES" | grep '\.py$' | head -1)

+ echo "$_DIFF_FILES" | grep -qE '(pyproject\.toml|setup\.cfg|requirements.*\.txt)' && NON_PY_WARNINGS="${NON_PY_WARNINGS}⚠ dependency changes detected — not reviewed; verify Python imports still resolve\n"

+ echo "$_DIFF_FILES" | grep -qE '(Dockerfile|docker-compose.*\.yml)' && NON_PY_WARNINGS="${NON_PY_WARNINGS}⚠ container config changes detected — not reviewed\n"

+ _PY_DIFF=$(echo "$_DIFF_FILES" | grep '\.py$')

+ # warnings print BEFORE the early exit — they must emit even when no Python file exists

+ [ -n "$NON_PY_WARNINGS" ] && printf "$NON_PY_WARNINGS"

+ If `$NON_PY_WARNINGS` non-empty: include in report header regardless of whether Python files exist. No Python files → the bash `exit 0` above is the enforceable stop (DMI skill — prose "stop" not executable).

+

+ - `codemap_available=true` → copy `$CODEMAP_CONTEXT_STAGE` to `$RUN_DIR/codemap-context.md` once `$RUN_DIR` exists (Step 2). Each dimension spawn prompt gets a literal block holding only ITS slice of the batch results (the whole block in every prompt re-bills every query result to consumers that never read it, and results for skipped agents are injected nowhere): qa unit → `uncovered` + `mock-rdeps` entries; doc unit → `undocumented` + `xrefs --broken` entries; sw-engineer → `rdeps` + `central` + `diff-impact`. Entries whose only consumer was skipped by classification are not injected anywhere:

+ <this agent's slice of $RUN_DIR/codemap-context.md>

+ Read `$_DEV_SHARED/codex-prepass.md` for the **skip/run criteria only** (small-diff skip, availability check) — do NOT use its dispatch line as the spawn prompt: its args name no output path, so Step 2's watch on `$RUN_DIR/codex.md` would idle ~2 min on a file that never appears and Step 6 would silently skip.

+ Dispatch — substitute `<TARGET>` and `<RUN_DIR>` with resolved literal values before the call (a `Skill()` dispatch gets no run-dir preamble, so an unexpanded `$RUN_DIR` reaches Codex as literal dollar-sign text): `Skill(skill="bridge:review", args="Read-only adversarial review of <TARGET>. Look for bugs, missed edge cases, incorrect logic, and inconsistencies with existing code patterns. Write findings to <RUN_DIR>/codex.md; do not apply fixes.")` (requires `bridge@borda-ai-rig`).

+ 2. **Unit grouping** (always): fold the survivors into spawn units per §Merged spawn units below — Agents 3+6 share one spawn, Agents 4+5 share one spawn, Agents 1 and 2 stay standalone. A unit exists when at least one of its dimensions survived stage 1; its prompt carries only the surviving dimensions' instructions.

+ 3. **Relevance ranking** (default only): rank the units by evidence — changed files and lines in their dimensions' territory, what the classification implies, what the structural context flagged (a merged unit ranks by its strongest surviving dimension) — and spawn the top `FANOUT_MAX` (3) units. With `--full` (`FANOUT_CAP=0`) skip this stage and spawn every unit of stage 2.

+ - **Merge before you split**: two dimensions whose files overlap go to one agent, not two — the fixed pairs below are the floor, not the ceiling.

+ ### Merged spawn units (shared pattern with oss:review — each spawn costs ~120,851 tok fixed overhead, so paired dimensions share one spawn)

+

+ - **Agents 3+6 = ONE `foundry:perf-optimizer` spawn** covering Performance + Architecture/API design — spawn when either dimension survives classification preselection; the prompt includes only the surviving dimensions' instructions (e.g. FIX with a public-signature change spawns this unit with only the Agent-6 dimension active).

+ - **Agents 4+5 = ONE `foundry:doc-scribe` spawn** covering Documentation + Linting — same rule.

+ - Agents 1 (sw-engineer, incl. security augmentation), 2 (qa-specialist), and 7 (challenger) stay standalone spawns.

+ - A merged spawn writes **one file per covered dimension**, each with its OWN full sections + Confidence block — never blended: `perf-optimizer.md` + `solution-architect.md`, or `doc-scribe.md` + `linting-expert.md`. Downstream contracts (consolidator filename list, Step-4 cross-validation "same type as origin", report sections) key on those files and stay unchanged.

+ - Merged-spawn envelope = JSON array, one element per dimension file, same per-element schema as the standard envelope. An element absent from the array ⇒ that dimension gets the ⏱ marker — never silently omitted.

+

+ Launch spawn units simultaneously with Agent tool (security augmentation folded into Agent 1 — not separate spawn; Agent 6 optional; Agents 3+6 and 4+5 launch as merged units per §Merged spawn units). Every agent prompt must begin with the [run-dir preamble (canonical)](#run-dir-preamble-canonical) and end with:

+ **Codemap-py context preamble (substituted by orchestrator)**: rehydrate `IFS= read -r codemap_available < "${TMPDIR:-/tmp}/dev-review-codemap-available-${CSID}" 2>/dev/null || codemap_available=false`. When `codemap_available=true`, each dimension spawn prompt is prefixed with its `## Structural Context (codemap-py, codemap_available=true)` slice from `$RUN_DIR/codemap-context.md` per the per-agent slicing rules in Step 1. Agents must read that block first and skip redundant Grep/Read on symbols already covered by codemap output. Block absent → fall back to current file-read behaviour. Challenger (Agent 7) unchanged.

+ **Agent 3 — foundry:perf-optimizer** (merged spawn with Agent 6 — see §Merged spawn units): Analyze performance issues. Algorithmic complexity, Python loops that should be NumPy/torch ops, repeated computation, unnecessary I/O. ML code: check DataLoader config, mixed precision. Prioritize by impact. Findings to `$RUN_DIR/perf-optimizer.md`.

+ **Agent 4 — foundry:doc-scribe** (merged spawn with Agent 5 — see §Merged spawn units): Check documentation completeness. Public APIs without docstrings, missing Google style sections, outdated README, CHANGELOG gaps. Verify examples run. `codemap_available=true`: read `undocumented` + `xrefs --broken` sections from codemap context block first — `undocumented` enumerates symbols missing docstrings; `xrefs --broken` enumerates stale Sphinx refs. Skip docstring-scan Reads on listed symbols; fall back to file reads only when codemap output empty for a symbol needed or when verifying a specific finding. Findings to `$RUN_DIR/doc-scribe.md`.

+ **Agent 5 — foundry:linting-expert** (dimension covered by the Agent 4 merged spawn — see §Merged spawn units): Static analysis audit. Check ruff and mypy pass. Type annotation gaps on public APIs, suppressed violations without explanation, missing pre-commit hooks. Flag mismatched target Python version. Findings to `$RUN_DIR/linting-expert.md`.

+ **Agent 6 — foundry:solution-architect (optional, for changes touching public API boundaries; dimension covered by the Agent 3 merged spawn — see §Merged spawn units, findings to `$RUN_DIR/solution-architect.md`)**: Target touches `__init__.py` exports, adds/modifies Protocols or ABCs, changes module structure, introduces new public classes, **or changes the signature of any already-exported public function — added/removed/renamed params, new flags** → evaluate API design quality, coupling impact, backward compatibility, and consistency of any added symbol (name, placement, signature, param/flag, return shape) with the existing API surface — naming conventions, module organization, sibling patterns (e.g. a new bool that duplicates an existing `kind=`/`mode=` discriminator, or a helper added where an equivalent already lives → flag, reuse/extend the existing home instead). Skip for purely internal (non-exported) implementation changes.

+ if [ ! -f "$_DEV_SHARED/foundry--cross-validation-protocol.md" ]; then

+ echo "⚠ foundry--cross-validation-protocol.md not found at $_DEV_SHARED — Step 4 skipped; critical findings are unverified. It ships in this plugin — reinstall develop@borda-ai-rig."

+ cat "$_DEV_SHARED/foundry--cross-validation-protocol.md"

+ **Skill-specific**: use **same agent type** that raised finding as verifier (e.g., foundry:sw-engineer verifies foundry:sw-engineer's critical finding). **Spawn cap: max 3 verifier agents** — critical/blocking findings > 3 → group into batches of ≤2 findings per verifier (same origin type per batch); note grouped IDs in rationale; each finding still gets its own verdict. Same cap as the shared protocol and oss:review — unbounded verifier fanout costs ~120,851 tok per critical finding.

+ echo "$_REVIEW_TEMPLATE"

+ Substitute the resolved literal path into the consolidator spawn prompt ("Read the report template at `<path>` with the Read tool — it defines the output structure"); do NOT cat the template into orchestrator context — that bills the same ~579 tok twice (once here, once inside the prompt).

+ - **Parallel agent cleanup**: after all spawn units complete, review `TaskList` — delete any tasks created by sub-agents (not by lead orchestrator). Sub-agent task creation unintended, can leave zombie tasks.