bigpowers 2.88.10 → 2.88.12

package/.zcode/skills/change-request/SKILL.md

@@ -4,8 +4,8 @@ model: sonnet
 description: "Add a new requirement or reorder epics by WSJF against specs/release-plan.yaml and epic capsule directories. Modes Add and Reorder. Use when a new requirement arrives mid-release or the plan needs prioritization."
 ---
 
-# story: e45s29
-# story: e20s01
+<!-- story: e45s29 -->
+<!-- story: e20s01 -->
 
 # Change Request
 

package/.zcode/skills/commit-message/SKILL.md

@@ -4,7 +4,7 @@ model: haiku
 description: "Reviews working-tree changes, then drafts a Conventional Commits title/body and states the semantic-release version bump a single such commit would imply. Also notes which defensive-code categories were touched. Use when the user wants to commit recent work, prepare a Conventional Commits message, or asks for semantic-release / semver-consistent messaging before git commit."
 ---
 
-# story: e82s02
+<!-- story: e82s02 -->
 
 # Commit Message
 > **HARD GATE** — **HARD GATE** — Commits must follow Conventional Commits spec (type(scope): description). Do NOT use vague messages like 'fix' or 'updates.' The message must explain the 'why,' not the 'what.'

package/.zcode/skills/compose-workflow/SKILL.md

@@ -4,9 +4,9 @@ model: sonnet
 description: "Chain multiple bigpowers skills into a custom workflow recipe saved in specs/. Use when a project repeats a non-standard skill sequence, or user wants a documented playbook beyond orchestrate-project modes."
 ---
 
-# story: e09s01
-# story: e45s27
-# story: e21s04
+<!-- story: e09s01 -->
+<!-- story: e45s27 -->
+<!-- story: e21s04 -->
 
 # Compose Workflow
 > **HARD GATE** — **HARD GATE** — Workflows are orchestration, not automation. Do NOT create workflows for tasks that should be single skills. Workflow complexity must be justified.

package/.zcode/skills/delegate-task/SKILL.md

@@ -4,7 +4,7 @@ model: sonnet
 description: "Delegate one complex task to a single subagent, review its work in two stages before merging back. Sequential — one agent at a time, with oversight. Use when a task is complex and requires careful review before the result is accepted. Distinct from dispatch-agents (no parallelism here; reviewer sees full diff before proceeding)."
 ---
 
-# story: e45s30
+<!-- story: e45s30 -->
 
 # Delegate Task
 > **HARD GATE** — **HARD GATE** — Delegated work must have clear success criteria and verification commands. The delegate must be able to verify completion independently.

package/.zcode/skills/dispatch-agents/SKILL.md

@@ -4,12 +4,12 @@ model: sonnet
 description: "Dispatch multiple subagents in parallel on independent tasks. No waiting between them — all run concurrently. Use when tasks are truly decoupled and speed matters. Distinct from delegate-task (concurrent here, no inter-task review gate)."
 ---
 
-# story: e09s04
-# story: e45s38
+<!-- story: e09s04 -->
+<!-- story: e45s38 -->
 <!-- story: e45s11 -->
 
 
-# story: e45s30
+<!-- story: e45s30 -->
 
 # Dispatch Agents
 > **HARD GATE** — **HARD GATE** — Agent work must be parallelizable and have explicit synchronization points. Do NOT dispatch work that has hidden dependencies between agents.

package/.zcode/skills/evolve-skill/SKILL.md

@@ -4,10 +4,10 @@ model: opus
 description: "Benchmark-gated skill evolution — consume bigpowers-benchmark report, propose plan-work change, edit skill via craft-skill, re-run benchmark, record ADR. Use when a skill underperforms on benchmark or stocktake finds systemic gap."
 ---
 
-# story: e23s03
+<!-- story: e23s03 -->
 
-# story: e09s01
-# story: e09s05
+<!-- story: e09s01 -->
+<!-- story: e09s05 -->
 
 # Evolve Skill
 

package/.zcode/skills/kickoff-branch/SKILL.md

@@ -4,9 +4,9 @@ model: haiku
 description: "Create an isolated Git worktree/branch or Jujutsu workspace, then verify a clean test baseline before code. Use when starting a feature or task."
 ---
 
-# story: e51s03
-# story: e20s03
-# story: e82s02
+<!-- story: e51s03 -->
+<!-- story: e20s03 -->
+<!-- story: e82s02 -->
 
 
 # Kickoff Branch

package/.zcode/skills/migrate-spec/SKILL.md

@@ -4,12 +4,12 @@ model: sonnet
 description: "Detect GSD, spec-kit, or BMAD spec artifacts and transform them into bigpowers YAML layout (state.yaml, release-plan.yaml, epics/, requirements/, plans/, ADRs). Use when migrating foreign spec docs."
 ---
 
-# story: e25s01
-# story: e25s02
-# story: e25s03
-# story: e25s04
-# story: e25s05
-# story: e25s06
+<!-- story: e25s01 -->
+<!-- story: e25s02 -->
+<!-- story: e25s03 -->
+<!-- story: e25s04 -->
+<!-- story: e25s05 -->
+<!-- story: e25s06 -->
 
 # Migrate Spec
 

package/.zcode/skills/plan-work/SKILL.md

@@ -4,7 +4,7 @@ model: opus
 description: "PLANNING SPINE STEP 3 of 3 — Plan the work: write detailed implementation tasks into the active epic capsule (specs/epics/eNN-slug/). Produces countable-story-format .md specs and runnable -tasks.yaml files. Use after slice-tasks (step 2). Not a substitute for scope-work (step 1) or slice-tasks (step 2)."
 ---
 
-# story: e45s29
+<!-- story: e45s29 -->
 
 # Plan Work
 

package/.zcode/skills/request-review/SKILL.md

@@ -4,12 +4,23 @@ model: opus
 description: "Dispatch a fresh reviewer agent with a clean context to critique the code after audit-code passes. The reviewer has no shared state with the coding agent and gives a genuine second opinion. Use after audit-code passes, before committing, or when user wants an independent code review."
 ---
 
-# story: e45s28
+<!-- story: e45s28 -->
 
 # Request Review
 
 Dispatch fresh reviewer agents with clean contexts. Reviewers have no shared state — they find what the coding agent missed.
 
+## Target (argument)
+
+If an argument was passed after `/request-review` (e.g. `/request-review abc1234` or `/request-review HEAD~3..HEAD`), use it as the diff target when building the reviewer brief:
+
+```bash
+git diff <arg>        # range: abc..def, HEAD~3..HEAD
+git show <arg>        # single commit: abc1234
+```
+
+If no argument was given, review uncommitted working-tree changes.
+
 **Distinct from `audit-code`:** `audit-code` is self-review (internal). This skill dispatches external agents.
 
 **Solo developer note:** Reviewer agents replace the human reviewer.

package/.zcode/skills/research-first/SKILL.md

@@ -4,8 +4,8 @@ model: sonnet
 description: "Look-before-build — search registries, repo, existing skills, and web for prior art before implementing. Appends Prior Art to the spec. Use after survey-context and before elaborate-spec, when adding dependencies, or when the task may already be solved."
 ---
 
-# story: e03s01
-# story: e21s02
+<!-- story: e03s01 -->
+<!-- story: e21s02 -->
 
 # Research First
 

package/.zcode/skills/run-planning/SKILL.md

@@ -4,7 +4,7 @@ model: sonnet
 description: "DISCOVER-PHASE ADVANCER — Drive the discover-phase checklist (specs/planning-status.yaml) through survey-context → scope-work → research-first → elaborate-spec → plan-release → slice-tasks. NOT a duplicate of plan-work or the planning spine; it orchestrates the pre-coding discover phase only."
 ---
 
-# story: e24s03
+<!-- story: e24s03 -->
 
 # Run Planning
 

package/.zcode/skills/scope-work/SKILL.md

@@ -4,8 +4,8 @@ model: sonnet
 description: "PLANNING SPINE STEP 1 of 3 — Scope the work: define what is in and out of scope and save as specs/product/SCOPE_LATEST.yaml. Use before slice-tasks or plan-release on any new initiative. Not a substitute for slice-tasks (step 2) or plan-work (step 3)."
 ---
 
-# story: e03s01
-# story: e24s02
+<!-- story: e03s01 -->
+<!-- story: e24s02 -->
 
 # Scope Work
 

package/.zcode/skills/search-skills/SKILL.md

@@ -4,8 +4,8 @@ model: haiku
 description: "Find the right bigpowers skill from natural-language intent using a local lexical index over SKILL.md frontmatter. Use when unsure which skill to invoke, or at start of research-first."
 ---
 
-# story: e09s01
-# story: e21s01
+<!-- story: e09s01 -->
+<!-- story: e21s01 -->
 
 # Search Skills
 

package/.zcode/skills/security-review/SKILL.md

@@ -4,16 +4,16 @@ model: sonnet
 description: "AI-powered security analysis of code changes — traces data flow, detects injection, auth bypass, secrets exposure, and unsafe deserialization across files. Use when reviewing pending changes, before release-branch, during verify-work Phase 5, during build-epic Step 0 threat modeling, or when the user says \"security review\" or \"scan for vulns\"."
 ---
 
-# story: e45s41
+<!-- story: e45s41 -->
 <!-- story: e45s18 -->
-# story: e26s01
-# story: e26s02
-# story: e26s03
-# story: e26s04
-# story: e26s05
-# story: e26s06
-# story: e26s07
-# story: e45s26
+<!-- story: e26s01 -->
+<!-- story: e26s02 -->
+<!-- story: e26s03 -->
+<!-- story: e26s04 -->
+<!-- story: e26s05 -->
+<!-- story: e26s06 -->
+<!-- story: e26s07 -->
+<!-- story: e45s26 -->
 
 # Security Review
 

package/.zcode/skills/seed-conventions/SKILL.md

@@ -4,12 +4,12 @@ model: sonnet
 description: "Generate CLAUDE.md and CONVENTIONS.md for a brand-new project through a brief interview, and create the specs/ directory with evolved bigpowers structure (product/, tech-architecture/, verifications/, epics/archive/). Entry point for greenfield projects. Use when starting a new project from scratch, when user asks to set up AI agent conventions, or when there is no CLAUDE.md yet."
 ---
 
-# story: e10s01
-# story: e47s02
-# story: e10s02
-# story: e51s02
-# story: e45s21
-# story: e79s03
+<!-- story: e10s01 -->
+<!-- story: e47s02 -->
+<!-- story: e10s02 -->
+<!-- story: e51s02 -->
+<!-- story: e45s21 -->
+<!-- story: e79s03 -->
 
 
 # Seed Conventions

package/.zcode/skills/session-state/SKILL.md

@@ -4,7 +4,7 @@ model: haiku
 description: "Track implementation decisions and progress in specs/state.yaml to prevent context rot. Use at the start of a session to load context, and whenever a significant decision is made or a milestone is reached."
 ---
 
-# story: e45s23
+<!-- story: e45s23 -->
 
 # Session State
 > **HARD GATE** — **HARD GATE** — Session state must be synchronized with git state. If state.yaml conflicts with the working tree, halt and ask for clarification. Do NOT assume state is correct.

package/.zcode/skills/slice-tasks/SKILL.md

@@ -4,7 +4,7 @@ model: sonnet
 description: "PLANNING SPINE STEP 2 of 3 — Slice the work: break a scoped PRD into vertical-slice stories in specs/epics/. Use after scope-work (step 1), before plan-work (step 3). Not a substitute for scope-work or plan-work."
 ---
 
-# story: e45s29
+<!-- story: e45s29 -->
 
 # Slice Tasks
 

package/.zcode/skills/stocktake-skills/SKILL.md

@@ -4,7 +4,7 @@ model: sonnet
 description: "Sequential subagent batch audit of the bigpowers skill catalog — Quick Scan (changed only) or Full (all skills). Use during sustain phase, before a major release, or when catalog drift is suspected."
 ---
 
-# story: e09s01
+<!-- story: e09s01 -->
 <!-- story: e45s12 -->
 <!-- story: e79s04 -->
 

package/.zcode/skills/survey-context/SKILL.md

@@ -4,7 +4,7 @@ model: haiku
 description: "Per-task context bootstrap — reads existing specs/ and tech-architecture docs to map the current lifecycle phase and suggest the next skill. Use at the start of any task, when returning after a break, or when unsure what to do next. For deriving a tech-stack doc from scratch, use map-codebase first."
 ---
 
-# story: e82s02
+<!-- story: e82s02 -->
 
 # Survey Context
 

package/CHANGELOG.md

@@ -1,3 +1,25 @@
+## [2.88.12](https://github.com/danielvm-git/bigpowers/compare/v2.88.11...v2.88.12) (2026-10-25)
+
+
+### Bug Fixes
+
+* **install,skills:** config bridge append, worktree TDD gate, story tags, audit args ([#158](https://github.com/danielvm-git/bigpowers/issues/158)) ([120a0d9](https://github.com/danielvm-git/bigpowers/commit/120a0d993df3af671c1e3acdeb78934830ee7ddd)), closes [#149](https://github.com/danielvm-git/bigpowers/issues/149) [#142](https://github.com/danielvm-git/bigpowers/issues/142) [#140](https://github.com/danielvm-git/bigpowers/issues/140) [#141](https://github.com/danielvm-git/bigpowers/issues/141)
+* live receipts page data + restore green CI on main ([#159](https://github.com/danielvm-git/bigpowers/issues/159)) ([6136322](https://github.com/danielvm-git/bigpowers/commit/6136322a0f436be22c96f4a7a005d4744de29c29)), closes [#158](https://github.com/danielvm-git/bigpowers/issues/158)
+
+## [2.88.11](https://github.com/danielvm-git/bigpowers/compare/v2.88.10...v2.88.11) (2026-10-25)
+
+
+### Bug Fixes
+
+* **guard:** resolve git branch per command target, not process cwd ([#154](https://github.com/danielvm-git/bigpowers/issues/154)) ([07c7498](https://github.com/danielvm-git/bigpowers/commit/07c749803c4d5d1c2499c01255fad138ce88e870))
+* **lockfile:** add missing slash in skill path + self-contained resolve_repo_root ([#156](https://github.com/danielvm-git/bigpowers/issues/156)) ([04bda15](https://github.com/danielvm-git/bigpowers/commit/04bda157195f39816cd388a2fb4b0829ca19b20f))
+* **omp-hooks:** inGitRepo bail-early, accept type!(scope) form, echo subject ([#157](https://github.com/danielvm-git/bigpowers/issues/157)) ([9e1b11a](https://github.com/danielvm-git/bigpowers/commit/9e1b11aaf4c63fa951c6916fabda13a836b56f2d)), closes [#147](https://github.com/danielvm-git/bigpowers/issues/147) [#151](https://github.com/danielvm-git/bigpowers/issues/151) [#151](https://github.com/danielvm-git/bigpowers/issues/151)
+
+
+### Performance Improvements
+
+* **sync:** render all skills in one bash process per adapter ([#155](https://github.com/danielvm-git/bigpowers/issues/155)) ([1b56463](https://github.com/danielvm-git/bigpowers/commit/1b564639771c39c3d4bc712f09912751a56ab8ee)), closes [#137](https://github.com/danielvm-git/bigpowers/issues/137)
+
 ## [2.88.10](https://github.com/danielvm-git/bigpowers/compare/v2.88.9...v2.88.10) (2026-10-25)
 
 

package/extensions/omp-hooks.ts

@@ -7,7 +7,7 @@ import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
 import { Type } from "typebox";
 import { execSync } from "node:child_process";
 import { readdirSync, readFileSync, statSync } from "node:fs";
-import { dirname, join } from "node:path";
+import { dirname, join, resolve } from "node:path";
 import { fileURLToPath } from "node:url";
 
 // ---------------------------------------------------------------------------
@@ -28,8 +28,10 @@ const DANGEROUS_PATTERNS = [
   /push\s+-f(?:\s|$)/,
 ];
 
+// Conventional Commits specifies `type(scope)!:`, but many repositories write
+// `type!(scope):`. Accept both; the leading type is what matters.
 const CONVENTIONAL_RE =
-  /^(feat|fix|docs|style|refactor|perf|test|build|ci|chore|revert)(\([^)]+\))?!?:\s.+/;
+  /^(feat|fix|docs|style|refactor|perf|test|build|ci|chore|revert)!?(?:\([^)]+\))?!?:\s.+/;
 
 const PROTECTED_BRANCH_RE = /(?:^|\s|[:])(?:main|master)(?:\s|$)/;
 const PROTECTED_BRANCHES = ["main", "master"] as const;
@@ -157,11 +159,57 @@ function pluginRoot(): string {
 // Git safety helpers
 // ---------------------------------------------------------------------------
 
-function currentBranch(): string {
+// Command-aware branch resolution: `git commit|push` in a bash command must
+// be judged against the repo it will actually run in, not the agent process
+// cwd (which is the session root — often a different repository).
+// Resolve the directory: the last `cd <path>` before it (within the command
+// prefix), overridden by the last `git -C <path>` anywhere in the command.
+// Fail-closed: unresolvable paths fall back to the process cwd — previous
+// behavior, over-blocks, never under-blocks.
+function commandGitCwd(command: string, upTo: number, fallback: string): string {
+  const head = command.slice(0, upTo);
+  const cdMatches = [
+    ...head.matchAll(/\bcd\s+(?:"([^"]+)"|'([^']+)'|([^&|;<(`$][^\s&|;<>]*))/g),
+  ];
+  const cMatches = [
+    ...command.matchAll(/\bgit\s+-C\s+(?:"([^"]+)"|'([^']+)'|([^&|;<(`$][^\s&|;<>]*))/g),
+  ];
+  const m = cMatches.length
+    ? cMatches[cMatches.length - 1]
+    : cdMatches.length
+      ? cdMatches[cdMatches.length - 1]
+      : null;
+  if (!m) return fallback;
+  const p = m[1] ?? m[2] ?? m[3];
+  if (!p || p.includes("$") || p.includes("`")) return fallback;
+  try {
+    const abs = resolve(fallback, p);
+    return statSync(abs).isDirectory() ? abs : fallback;
+  } catch {
+    return fallback;
+  }
+}
+
+function inGitRepo(cwd?: string): boolean {
+  try {
+    execSync("git rev-parse --git-dir", {
+      cwd,
+      encoding: "utf-8",
+      timeout: 3000,
+      stdio: "ignore",
+    });
+    return true;
+  } catch {
+    return false;
+  }
+}
+
+function currentBranch(cwd: string): string {
   try {
     return execSync("git rev-parse --abbrev-ref HEAD", {
       encoding: "utf-8",
       timeout: 3000,
+      cwd,
     }).trim();
   } catch {
     return "unknown";
@@ -367,9 +415,11 @@ export default function bigpowers(pi: ExtensionAPI) {
       }
     }
 
-    const branch = currentBranch();
+    const sessionRoot = process.cwd();
+    if (!inGitRepo(sessionRoot)) return;
+    const pushBranch = currentBranch(commandGitCwd(command, command.length, sessionRoot));
 
-    if (/git\s+push/.test(command) && protectedPushTarget(command, branch)) {
+    if (/git\s+push/.test(command) && protectedPushTarget(command, pushBranch)) {
       return {
         block: true,
         reason:
@@ -378,10 +428,16 @@ export default function bigpowers(pi: ExtensionAPI) {
     }
 
     if (/git\s+commit/.test(command)) {
-      if ((PROTECTED_BRANCHES as readonly string[]).includes(branch)) {
+      const commitBranches = [...command.matchAll(/\bgit\s+commit\b/g)].map((m) =>
+        currentBranch(commandGitCwd(command, m.index ?? 0, sessionRoot)),
+      );
+      const hit = commitBranches.find((b) =>
+        (PROTECTED_BRANCHES as readonly string[]).includes(b),
+      );
+      if (hit) {
         return {
           block: true,
-          reason: `BLOCKED: Direct commits to '${branch}' are forbidden. Use a feature branch.`,
+          reason: `BLOCKED: Direct commits to '${hit}' are forbidden. Use a feature branch.`,
         };
       }
       const msg = extractCommitMessage(command);
@@ -389,8 +445,7 @@ export default function bigpowers(pi: ExtensionAPI) {
         if (!CONVENTIONAL_RE.test(msg)) {
           return {
             block: true,
-            reason:
-              "BLOCKED: Commit message must follow Conventional Commits: <type>(<scope>): <subject>.",
+            reason: `BLOCKED: Commit subject '${msg.split("\n")[0]}' must follow Conventional Commits: <type>(<scope>): <subject>.`,
           };
         }
         if (msg.split("\n")[0].length > 72) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "bigpowers",
-  "version": "2.88.10",
+  "version": "2.88.12",
   "description": "73 agent skills synthesizing 17 years of software engineering discipline into a prescriptive methodology for solo developers",
   "main": "index.js",
   "scripts": {

package/scripts/lib/adapter-render-all.sh

@@ -0,0 +1,74 @@
+#!/usr/bin/env bash
+# story: #138
+# adapter-render-all.sh — render every skill through one adapter in a single
+# bash process.
+#
+# srp-engine.py --all used to spawn one `bash` (plus three `jq`) per
+# skill×adapter — ~1,539 bash and ~4,600 jq processes from one Python process.
+# On constrained Windows sessions that exhausts process-creation resources: the
+# run dies with STATUS_DLL_INIT_FAILED (0xC0000142) and then hangs on orphaned
+# children. Sourcing the adapter once and calling render_skill() in a loop keeps
+# the exact same rendering code while collapsing the process count to one per
+# adapter and removing the per-render jq calls.
+#
+# Input is a NUL-delimited stream written by srp-engine.py, five fields per
+# skill in order — name, description, model, body, pi body — each NUL-terminated
+# so field content may contain newlines.
+#
+# Adapters without a stdin entrypoint (continue, omp, zed) define render_skill
+# but were never driven by srp-engine; they are skipped so behaviour is
+# unchanged.
+#
+# Usage: adapter-render-all.sh <path-to-adapter.sh> <ir-stream-file>
+set -u
+
+adapter="$1"
+stream="$2"
+
+if [[ ! -f "$adapter" ]]; then
+  echo "adapter-render-all: adapter not found: $adapter" >&2
+  exit 1
+fi
+if [[ ! -f "$stream" ]]; then
+  echo "adapter-render-all: IR stream not found: $stream" >&2
+  exit 1
+fi
+
+# A spawned adapter with no stdin entrypoint is a no-op; match that.
+if ! grep -q '\[\[ ! -t 0 \]\]' "$adapter"; then
+  exit 0
+fi
+
+adapter_id="$(basename "$adapter" .sh)"
+
+# Sourcing the adapter runs its stdin entrypoint; point stdin at /dev/null so
+# `[[ ! -t 0 ]]` is true but `cat` yields nothing and the entrypoint no-ops.
+# shellcheck disable=SC1090
+source "$adapter" < /dev/null
+
+if ! declare -f render_skill >/dev/null 2>&1; then
+  exit 0
+fi
+
+exec 3< "$stream"
+status=0
+while IFS= read -r -d '' IR_NAME <&3; do
+  IFS= read -r -d '' IR_DESCRIPTION <&3
+  IFS= read -r -d '' IR_MODEL <&3
+  IFS= read -r -d '' IR_BODY <&3
+  IFS= read -r -d '' IR_BODY_PI <&3
+
+  IR_DESC_ESCAPED=$(echo "$IR_DESCRIPTION" | sed 's/\\/\\\\/g; s/"/\\"/g')
+  if [[ "$adapter_id" == "pi" ]]; then
+    IR_BODY_SKILL="$IR_BODY_PI"
+  else
+    IR_BODY_SKILL="$IR_BODY"
+  fi
+
+  if ! render_skill; then
+    status=1
+    break
+  fi
+done
+exec 3<&-
+exit "$status"

package/scripts/lib/context-wire.sh

@@ -34,7 +34,13 @@ wire_config_bridge() {
   if [[ -f "$bridge_file" ]] && grep -q "^${bridge_key}:" "$bridge_file" 2>/dev/null; then
     return 0
   fi
-  printf '%s: %s\n' "$bridge_key" "$agents_path" > "$bridge_file"
+  if [[ -f "$bridge_file" ]]; then
+    cp "$bridge_file" "${bridge_file}.bak"
+    printf '%s: %s\n' "$bridge_key" "$agents_path" >> "$bridge_file"
+    echo "install: appended ${bridge_key} to ${bridge_file} (existing content preserved; backup at ${bridge_file}.bak)"
+  else
+    printf '%s: %s\n' "$bridge_key" "$agents_path" > "$bridge_file"
+  fi
 }
 
 wire_context_mode() {

package/scripts/lib/install-helpers.js

@@ -142,6 +142,10 @@ function bridgeHermesConfig(configPath) {
   if (fs.existsSync(configPath)) {
     const existing = fs.readFileSync(configPath, 'utf8');
     if (/^instructions:/m.test(existing)) return;
+    fs.copyFileSync(configPath, configPath + '.bak');
+    fs.appendFileSync(configPath, 'instructions: AGENTS.md\n');
+    console.log(`install: appended instructions to ${configPath} (existing content preserved; backup at ${configPath}.bak)`);
+    return;
   }
   fs.writeFileSync(configPath, 'instructions: AGENTS.md\n');
 }

package/scripts/lib/srp-engine.py

@@ -7,6 +7,7 @@ import json
 import glob
 import shutil
 import subprocess
+import tempfile
 
 # Ensure scripts/lib is on the path so sibling modules (link_utils, and the
 # simple_yaml fallback below) resolve regardless of cwd.
@@ -231,6 +232,50 @@ def dispatch_to_adapter(skill_data, target, repo_root):
         print(f"Error: adapter {target} failed with exit code {proc.returncode}", file=sys.stderr)
         sys.exit(proc.returncode)
 
+def _write_ir_stream(records):
+    fd, stream_path = tempfile.mkstemp(prefix="bigpowers-ir-", suffix=".bin")
+    with os.fdopen(fd, "wb") as stream:
+        for data in records:
+            pi_body = rewrite_links_for_pi(data['body'], data['name'])
+            for field in (data['name'], data['description'], data['model'], data['body'], pi_body):
+                stream.write(field.encode("utf-8"))
+                stream.write(b"\x00")
+    return stream_path
+
+
+def render_all(records, adapters, repo_root):
+    """Render every skill through each adapter with one bash process per adapter.
+
+    The previous path spawned one bash (plus three jq) per skill×adapter — about
+    1,539 bash and 4,600 jq processes from a single Python process. On constrained
+    Windows sessions that exhausts process-creation resources, so the run dies
+    with STATUS_DLL_INIT_FAILED (0xC0000142) and then hangs on orphaned children.
+    Writing the IR once and letting a small bash driver source the adapter keeps
+    the same rendering code while dropping to one process per adapter.
+    """
+    if not records:
+        return
+
+    driver = os.path.join(repo_root, "scripts", "lib", "adapter-render-all.sh")
+    bash = resolve_bash()
+    stream_path = _write_ir_stream(records)
+    try:
+        for adapter in adapters:
+            adapter_path = os.path.join(repo_root, "scripts", "adapters", f"{adapter}.sh")
+            if not os.path.isfile(adapter_path):
+                print(f"Error: adapter not found: {adapter_path}", file=sys.stderr)
+                sys.exit(1)
+            proc = subprocess.run([bash, driver, adapter_path, stream_path])
+            if proc.returncode != 0:
+                print(f"Error: adapter {adapter} failed with exit code {proc.returncode}", file=sys.stderr)
+                sys.exit(1)
+    finally:
+        try:
+            os.unlink(stream_path)
+        except OSError:
+            pass
+
+
 def main():
     repo_root = resolve_repo_root()
 
@@ -241,19 +286,17 @@ def main():
 
         okf_wiki_skills = os.environ.get("OKF_WIKI_SKILLS", os.path.join(repo_root, "specs", "skills-wiki", "skills"))
 
+        records = []
         for skill_md in skills:
             skill_data = parse_skill(skill_md)
             if not skill_data['name']:
                 continue
+            records.append(skill_data)
 
-            for adapter in adapters:
-                data = skill_data
-                if adapter == "pi":
-                    data = dict(skill_data)
-                    data["body_pi_skill"] = rewrite_links_for_pi(skill_data['body'], skill_data['name'])
-                dispatch_to_adapter(data, adapter, repo_root)
+        render_all(records, adapters, repo_root)
 
-            if okf_mode:
+        if okf_mode:
+            for skill_data in records:
                 render_okf_concept(skill_data, okf_wiki_skills)
         return
 

package/scripts/regenerate-lockfile.sh

@@ -3,6 +3,7 @@
 # regenerate-lockfile.sh — Scan all SKILL.md files and regenerate skills-lock.json
 set -euo pipefail
 source "$(dirname "${BASH_SOURCE[0]}")/lib/skill-common.sh"
+resolve_repo_root
 LOCKFILE="$REPO_ROOT/skills-lock.json"
 records=()
 while IFS= read -r skill_dir; do
@@ -12,7 +13,7 @@ while IFS= read -r skill_dir; do
   description="$_PF_DESCRIPTION"
   if command -v sha256sum &>/dev/null; then sha256=$(sha256sum "$skill_md" | cut -c1-16)
   else sha256=$(shasum -a 256 "$skill_md" | cut -c1-16); fi
-  relpath="${skill_dir#$REPO_ROOT/}SKILL.md"
+  relpath="${skill_dir#$REPO_ROOT/}/SKILL.md"
   records+=("$(jq -n --arg name "$name" --arg desc "$description" --arg sha "$sha256" --arg path "$relpath" \
     '{($name): {description: $desc, sha256: $sha, path: $path}}')")
 done < <(iterate_skills)