@blamejs/exceptd-skills 0.12.41 → 0.13.0

package/CHANGELOG.md

@@ -1,5 +1,96 @@
 # Changelog
 
+## 0.13.0 — 2026-05-17
+
+Minor release. Breaking-change bundle for the v0.10.x legacy-verb removal that has been deprecation-bannered since v0.11.0; envelope harmonization across every JSON-emitting verb; 4 new playbooks expanding the canonical set to 20; engine hardening (factor-shape validation, cache invalidation, fsync-on-rename, deterministic SBOM); schema reverse fields on ATLAS, ATT&CK, and the playbook chain.
+
+### Breaking changes — migration required
+
+**Five v0.10.x legacy verbs hard-removed: `plan`, `govern`, `direct`, `look`, `ingest`.** They were deprecation-bannered since v0.11.0 and slated-for-removal-in-v0.13 since v0.12.0. Operators on v0.10.x → v0.13.0 now get a structured `ok:false` refusal with the v0.11+ replacement command. Each removal is a pure rename — same underlying capability is reachable via the replacement. Refusal body shape:
+
+```json
+{
+  "ok": false,
+  "error": "'plan' was removed in v0.13.0. Use `exceptd brief --all` instead.",
+  "verb": "plan",
+  "removed_in": "0.13.0",
+  "replacement": "brief --all",
+  "deprecation_history": "Deprecated in v0.11.0 ... removed in v0.13.0."
+}
+```
+
+Replacements:
+- `exceptd plan` → `exceptd brief --all`
+- `exceptd govern <pb>` → `exceptd brief <pb> --phase govern`
+- `exceptd direct <pb>` → `exceptd brief <pb> --phase direct`
+- `exceptd look <pb>` → `exceptd brief <pb> --phase look`
+- `exceptd ingest <args>` → `exceptd run <args>`
+
+`reattest` and `list-attestations` were also deprecation-bannered but are PRESERVED — they remain canonical short-form routings of `attest diff` / `attest list` and stay functional.
+
+The deprecation-banner + tempdir-marker mechanism (added v0.11.0, persisted via `EXCEPTD_DEPRECATION_SHOWN` env var + `exceptd-deprecation-shown-v<X.Y.Z>` tempdir marker) is removed. Pre-v0.13 scripts that pinned the banner shape should remove those assertions.
+
+**Orchestrator exit-code class change: usage errors exit 1 (`GENERIC_FAILURE`), not 2 (`DETECTED_ESCALATE`).** Affected verbs: `framework-gap` (missing args), `report <format>` (unknown format), `validate-cves` / `validate-rfcs` (catalog-read failure), `watchlist` (manifest-read failure), `skill <name>` (skill-not-found). Pre-v0.13 these exited 2, colliding with the canonical CI contract where exit 2 means "verb ran and detected an escalation-worthy finding." CI gates wired to branch on exit 2 will need to also accept exit 1 for these verbs, OR pre-validate inputs before invocation.
+
+**Envelope harmonization: every JSON-emitting verb now carries top-level `ok` and (where applicable) `verb`.** Pre-v0.13 `brief --all`, `watchlist`, `ci`, `doctor`, `discover`, `attest show`, `attest export`, and `cmdRunMulti` omitted one or both fields inconsistently. `emit()` now defaults `ok: true` when not set (symmetric to the existing `ok: false → exit 1` fallback), and per-verb call sites set `verb: "..."` explicitly. Consumers that parsed bodies for the absence of these fields will break; consumers reading specific known fields are unaffected.
+
+**Orchestrator `ok:false` bodies now land on stdout (not stderr).** Aligns with the bin/exceptd.js convention so a single consumer can parse the verb's envelope without splitting across two streams. Advisory text (`[verb] hint: ...`) still goes to stderr.
+
+### Security
+
+**`lib/sign.js` `generateKeypair()` ACL-hardening status surfaces in CLI output.** `restrictWindowsAcl()` now returns a boolean; the verdict line announces `Windows ACL hardened: yes|NO` rather than silently warning.
+
+**`lib/cve-curation.js` + `lib/refresh-external.js` `writeJsonAtomic()` fsync before rename.** Pre-v0.13 a power loss between the tmp-write and the rename could leave the renamed destination zero-length / partial. The open + write + fsync + close + rename idiom closes the durability gap on both atomic-write helpers.
+
+### Features
+
+**4 new playbooks expand the canonical set to 20.**
+- `webhook-callback-abuse` — OAuth callback hijack, inbound-webhook signature validation, Slack/Teams/Discord webhook leakage, the Snowflake-class long-lived-callback-token-in-CI-log pattern.
+- `cicd-pipeline-compromise` — self-hosted runner takeover, workflow-injection (the `${{ inputs.* }}` class), third-party Action SHA-pin discipline, OIDC trust-policy abuse, runner-scoped signing keys. Distinct from `sbom` (package-registry supply chain).
+- `identity-sso-compromise` — in-progress IdP-plane detection (Salt Typhoon / Scattered Spider / Okta-2023 / golden-SAML / PRT theft patterns). Detect-side counterpart to the existing `idp-incident` (response playbook).
+- `llm-tool-use-exfil` — agentic-AI tool abuse via prompt injection. Auto-approve-on-high-impact-tool, instruction-coercion grammar in tool responses, unprompted tool chains, credential-shadow in tool args. Distinct from `dlp-exfiltration` (enterprise DLP) and `mcp` (install-time tool trust).
+
+Each new playbook carries `threat_currency_score >= 90`, full air-gap alternatives on look artifacts, substantive theater_test blocks on framework gaps, and `feeds_into[]` chains into the existing playbook set.
+
+**Schema reverse fields populated across 3 catalogs + playbooks.**
+- ATLAS TTPs now carry `cve_refs[]` (11 entries populated with 21 back-edges from `cve.atlas_refs`).
+- ATT&CK techniques now carry `cve_refs[]` (20 entries populated with 56 back-edges from `cve.attack_refs`).
+- Every playbook now carries `_meta.fed_by[]` (11 playbooks populated with 54 back-edges from `_meta.feeds_into[].playbook_id`). Operators reading a playbook can see what chains INTO it without grepping every other playbook.
+
+`scripts/refresh-reverse-refs.js` extends to 8 reverse-direction passes (4 manifest-driven, 2 CVE-driven, 2 catalog-back-edge, 1 playbook-back-edge); `npm run refresh-reverse-refs` rebuilds the full set in one pass.
+
+**`lib/scoring.js validate()` refuses mixed Shape A / Shape B factor sets.** The catalog historically stored `rwep_factors` in two distinct shapes (raw booleans + post-weight integers). Mixing shapes inside one entry silently broke the sum invariant — a CVE with `cisa_kev: true, blast_radius: 30` reported rwep 30 instead of the operator-intended 55. The new `detectFactorShape()` helper detects mixed entries and emits a structured error pointing at the affected CVE id.
+
+**`lib/cross-ref-api.js` cache invalidation uses (mtime, size) tier.** Pre-v0.13 cache invalidation was mtime-only; on filesystems with 1-2s mtime granularity (FAT32, HFS+ pre-APFS, NFSv3, Docker bind-mounts that proxy mtime) a rapid refresh-then-reload within the same second served stale data. Adding `size` catches every content change that affects byte count.
+
+**`lib/lint-skills.js` enforces `last_threat_review` staleness gate.** Warn at >180 days; hard fail at >365 days. Operators with stale skills get a structured warning naming the affected file + the exact day count; year-stale skills fail the lint outright. The forcing function for Hard Rule #8 (which was policy-only pre-v0.13).
+
+**`scripts/refresh-sbom.js` produces a deterministic bundle.** `metadata.timestamp` and `serialNumber` are derived from the content-hash seed (`<name>@<version>@<bundleSha>`) instead of wall-clock. Identical content → identical SBOM across re-runs. The SBOM-currency predeploy gate can now rely on byte-identity for the no-change case.
+
+**`exceptd doctor --fix` second remediation branch: post-rotate stale signatures.** Continues the v0.12.41 fix. When the private key IS present but the signatures check fails (the `generate-keypair --rotate` followup state), `doctor --fix` runs `sign-all` to re-sign skills + manifest against the current keypair. Without this branch the rotation flow would converge to a broken-but-not-self-healing state.
+
+### Bugs
+
+**3 catalogs got `last_threat_review` fields backfilled.** `exploit-availability.json`, `global-frameworks.json`, `zeroday-lessons.json` carried `last_updated` but lacked the threat-review timestamp the other 8 catalogs use. All 11 now follow the same shape.
+
+**`active_exploitation` field vocabulary now declared in `cve-catalog.json._meta`.** Pre-v0.13 the field accepted free-form values; 10 entries used `"unknown"` which wasn't documented. The new `_meta.active_exploitation_vocabulary` block enumerates `confirmed | suspected | theoretical | none | unknown` with per-value definitions.
+
+**4 CVEs flipped `_draft: false` (verified).** `CVE-2024-3094` (xz-utils backdoor), `CVE-2024-21626` (Leaky Vessels), `CVE-2026-42945` (NGINX Rift), `MAL-2026-TANSTACK-MINI`. 1 quarantined (`MAL-2026-ANTHROPIC-MCP-STDIO` — duplicate of `CVE-2026-30623`). The remaining 15 draft CVEs are now marked with a structured `_draft_reason` ("blocked on missing zeroday-lessons entry" in all 14 cases except the GTIG embargoed placeholder).
+
+**Hard Rule #5 regional-framework backfill on 7 skills.** `policy-exception-gen`, `compliance-theater`, `exploit-scoring`, `ai-c2-detection`, `ai-attack-surface`, `mcp-agent-trust`, `api-security` previously cited NIST without one or more of EU/UK/AU/ISO equivalents. Each now carries substantive references to NIS2/DORA/EU AI Act, NCSC CAF, ASD ISM/Essential 8, and ISO 27001:2022 controls as appropriate.
+
+**39 of 42 skills got `data_deps` arrays regenerated** from body content references. Pre-v0.13 the array drifted whenever a skill body added or removed a `data/<file>.json` reference without the frontmatter being updated. `api-security` and `email-security-anti-phishing` ended up with empty `data_deps` — their bodies reference no catalog file by name; flagged for v0.14 body-content review.
+
+**`scripts/refresh-reverse-refs.js` orphan detection caught 8 framework-gap forward-orphan references introduced by the Hard Rule #5 backfill.** `NIST-800-53-SC-39`, `ISO-27001-2022-A.8.22`, `CIS-Kubernetes-Benchmark-5.7`, `NIST-800-218-SSDF-PW.4`, `NIST-800-53-SR-3`, `SLSA-v1.0-Source-L3`, `NIST-AI-RMF-MAP-3.4`, `OWASP-Top-10-2021-A06`. These are real framework controls cited by the new skill content but absent from the gap catalog; tracked for v0.14 gap-catalog expansion.
+
+### Internal
+
+- `lib/exit-codes.js` exports `safeExit(code)` — sets `process.exitCode` without calling `process.exit()`. Dispatch surface (`bin/exceptd.js`, `orchestrator/index.js`) converted from `process.exit(N)` to `safeExit(EXIT_CODES.X)` for non-zero codes; `tests/safe-exit-grep.test.js` refuses regressions.
+- `validate-playbooks` predeploy gate flipped from informational to required. 20/20 playbooks validate cleanly.
+- 7 new pinning tests in `tests/v0_12_41-fixes.test.js`, `tests/safe-exit-grep.test.js`, `tests/atlas-version-canonical.test.js`, `tests/operator-leak-grep.test.js`, `tests/verify-shipped-tarball-wrapper.test.js`.
+- Test suite: 1179 total, 1173 pass, 6 skipped (POSIX-only / no-privkey gates), 0 fail.
+- `release.yml` publish-job split (id-token:write vs contents:write separation) and `refresh.yml` split-checkout pattern remain in v0.14 backlog; they're workflow-security hardening with no operator-facing surface change.
+
 ## 0.12.41 — 2026-05-17
 
 Cross-domain hygiene pass: signature-regression class fix, sidecar hardening, attestation UX, ATLAS pin reconciliation, operator-narrative scrub, and structural test pins to lock the class fixes against future drift.

package/bin/exceptd.js

@@ -207,31 +207,46 @@ function suggestVerb(cmd, known) {
 // v0.11.0 introduces: brief (collapses plan/govern/direct/look), discover (scan + dispatch),
 // doctor (currency + verify + validate-cves + validate-rfcs), ci (CI gate),
 // ai-run (streaming JSONL), ask (plain-English routing).
+//
+// v0.13.0 removed the v0.10.x phase-name aliases (plan, govern, direct,
+// look, ingest). They were deprecation-bannered since v0.11.0 and
+// slated-for-removal-in-v0.13 since v0.12.0; v0.13 honors that contract.
+// REMOVED_VERBS below carries the rename map for operator-facing refusal
+// hints. `reattest` and `list-attestations` are preserved as canonical
+// routings — they're short forms of `attest diff` / `attest list` that
+// remain operationally useful and have substantial test coverage.
 const PLAYBOOK_VERBS = new Set([
-  // v0.11.0 canonical surface:
   "brief", "run", "ai-run", "attest", "discover", "doctor", "ci", "ask",
   "verify-attestation", "run-all", "lint",
-  // v0.10.x legacy verbs — kept as aliases with deprecation banner, scheduled for removal in v0.13:
-  "plan", "govern", "direct", "look", "ingest", "reattest", "list-attestations",
+  "reattest", "list-attestations",
 ]);
 
-// Map legacy verb names to their v0.11.0 replacement so the dispatcher can
-// emit a single deprecation banner per session.
-const LEGACY_VERB_REPLACEMENTS = {
+// v0.13.0: hard-removed legacy verbs. The dispatcher refuses the verb
+// with an actionable replacement hint instead of routing it. Pre-v0.13
+// these were soft-deprecated (banner + still functional); v0.13 removes
+// the routing entirely. Operators upgrading from v0.10.x → v0.13 see
+// the same hint that the deprecation banner previously surfaced, but
+// non-zero exit so scripts noticing pinned-name use fail loudly instead
+// of silently invoking the alias.
+const REMOVED_VERBS = {
   plan: "brief --all",
   govern: "brief <pb> --phase govern",
   direct: "brief <pb> --phase direct",
   look: "brief <pb> --phase look",
   ingest: "run",
-  reattest: "attest diff",
-  "list-attestations": "attest list",
+};
+
+// Renamed but functionally-routed verbs (orchestrator-side dispatch still
+// handles them as of v0.13). Distinct from REMOVED_VERBS — these aren't
+// refused; they're just a soft hint that the rename happened. No banner
+// is emitted post-v0.13.
+const RENAMED_VERBS_HINT = {
   scan: "discover --scan-only",
   dispatch: "discover",
   currency: "doctor --currency",
   verify: "doctor --signatures",
   "validate-cves": "doctor --cves",
   "validate-rfcs": "doctor --rfcs",
-  watchlist: "watch",
   prefetch: "refresh --no-network",
   "build-indexes": "refresh --indexes-only",
 };
@@ -498,40 +513,23 @@ function main() {
     process.exit(0);
   }
 
-  // v0.12.8: emit the deprecation banner BEFORE branching on PLAYBOOK_VERBS
-  // so that legacy aliases routed through STANDALONE_VERBS or the orchestrator
-  // (scan, dispatch, currency, verify, validate-cves, validate-rfcs,
-  // watchlist, prefetch, build-indexes) also surface the rename.
-  // Previously the banner only fired for PLAYBOOK_VERBS-resident aliases
-  // (plan, govern, direct, look, ingest, reattest, list-attestations).
-  if (LEGACY_VERB_REPLACEMENTS[cmd] && !process.env.EXCEPTD_DEPRECATION_SHOWN) {
-    const ver = readPkgVersion();
-    // Cycle 11 F7 (v0.12.32): persist the suppression across invocations via
-    // an OS-tempdir marker keyed by exceptd version. Pre-fix the env-var
-    // guard reset every fresh node process so operators saw the same banner
-    // on every `exceptd plan` invocation, even after they'd already read it.
-    // Per-version key means a new version (legitimate new content) shows the
-    // banner once; subsequent runs within the same version stay quiet. The
-    // explicit EXCEPTD_DEPRECATION_SHOWN=1 env-var opt-out still suppresses
-    // even the first display, matching the documented contract.
-    const markerDir = require("os").tmpdir();
-    const markerFile = path.join(markerDir, `exceptd-deprecation-shown-v${ver}`);
-    let alreadyShown = false;
-    try { alreadyShown = fs.existsSync(markerFile); } catch { /* tmpdir unwritable; degrade to per-process */ }
-    if (!alreadyShown) {
-      const haveBrief = ver !== "unknown" && ver.match(/^(\d+)\.(\d+)/) && (parseInt(RegExp.$1, 10) > 0 || parseInt(RegExp.$2, 10) >= 11);
-      process.stderr.write(
-        `[exceptd] DEPRECATION: \`${cmd}\` is a v0.10.x verb. ` +
-        (haveBrief
-          ? `Prefer \`${LEGACY_VERB_REPLACEMENTS[cmd]}\` (available in this install, v${ver}). `
-          : `Upgrade to v0.11.0+ then use \`${LEGACY_VERB_REPLACEMENTS[cmd]}\` (currently installed: v${ver}). `) +
-        `Legacy verbs remain functional through this release; they will be removed in v0.13. ` +
-        `This banner shows once per exceptd version per host (re-shown on upgrade). Permanent suppress: export EXCEPTD_DEPRECATION_SHOWN=1.\n`
-      );
-      try { fs.writeFileSync(markerFile, `shown_at=${new Date().toISOString()}\nversion=${ver}\n`, { mode: 0o600 }); }
-      catch { /* tmpdir unwritable; the env-var guard below keeps the per-process suppression intact */ }
-    }
-    process.env.EXCEPTD_DEPRECATION_SHOWN = "1";
+  // v0.13.0: hard-refuse the v0.10.x legacy verbs that were
+  // deprecation-bannered since v0.11.0. Pre-v0.13 these silently routed
+  // to their v0.11+ replacements with a soft banner; v0.13 honors the
+  // long-advertised removal. Operators upgrading from v0.10.x get a
+  // structured error with the replacement command, suitable for
+  // grep / scripted handling.
+  if (REMOVED_VERBS[cmd]) {
+    emitError(
+      `'${cmd}' was removed in v0.13.0. Use \`exceptd ${REMOVED_VERBS[cmd]}\` instead.`,
+      {
+        verb: cmd,
+        removed_in: "0.13.0",
+        replacement: REMOVED_VERBS[cmd],
+        deprecation_history: "Deprecated in v0.11.0 with a soft banner; slated-for-removal-in-v0.13 announced in v0.12.0; removed in v0.13.0.",
+      }
+    );
+    return;
   }
 
   // Seven-phase playbook verbs run in-process — they emit JSON to stdout
@@ -714,6 +712,16 @@ function emit(obj, pretty, humanRenderer) {
   if (obj && obj.ok === false && !process.exitCode) {
     process.exitCode = EXIT_CODES.GENERIC_FAILURE;
   }
+  // v0.13.0 envelope harmonization: every emitted body has a top-level
+  // `ok` field — defaults to true when not set, matching the symmetric
+  // ok:false → exitCode=1 fallback above. Consumers that parse stdout
+  // can now assume the envelope shape regardless of which verb produced
+  // the body. Per-site `verb: "<name>"` is set at the call site; this
+  // helper guarantees the `ok` field's presence but does not synthesize
+  // verb (the caller knows its own name).
+  if (obj && typeof obj === 'object' && !('ok' in obj)) {
+    obj = { ok: true, ...obj };
+  }
   const wantJson = !!global.__exceptdWantJson || !!process.env.EXCEPTD_RAW_JSON;
   if (humanRenderer && !wantJson && !pretty) {
     process.stdout.write(humanRenderer(obj) + "\n");
@@ -4544,7 +4552,7 @@ function cmdAttest(runner, args, runOpts, pretty) {
   }
 
   if (subverb === "show") {
-    emit({ session_id: sessionId, attestations, attestation_replays: replays }, pretty);
+    emit({ verb: "attest show", session_id: sessionId, attestations, attestation_replays: replays }, pretty);
     return;
   }
 

package/data/_indexes/_meta.json

@@ -1,63 +1,63 @@
 {
   "schema_version": "1.1.0",
-  "generated_at": "2026-05-17T20:21:35.517Z",
+  "generated_at": "2026-05-17T23:47:25.094Z",
   "generator": "scripts/build-indexes.js",
   "source_count": 54,
   "source_hashes": {
-    "manifest.json": "adcee350ac3e4ffc9a6edd5003735df10503bffcd7474e298d1ecddb1e948224",
-    "data/atlas-ttps.json": "259e76e4252c7a56c17bbe96982a5e37ac89131c2d37a547fe38d64dcacfd763",
-    "data/attack-techniques.json": "51f60819aef36e960fd768e44dcc725e137781534fbbb028e5ef6baa21defa1d",
-    "data/cve-catalog.json": "415384660f879d26491a612e579943092135acb676cbb95327f309c921ff69ed",
-    "data/cwe-catalog.json": "6e7349a0fac39bdf9c4cb4598e101e51400f67d64c5d653bbca462f28bc1a0cb",
+    "manifest.json": "e4db72d2b307adb316fad1af15322392587c50c4de5b7592c015b93631f6f26b",
+    "data/atlas-ttps.json": "1301e6d053be91ac41ec38e64c72ac726e478b7fd6482884c3e2219c8fa16173",
+    "data/attack-techniques.json": "4afc226cc7359771a54aab1a1ff53c5530d1ef54eabc1d2fbd3b1caeda2f64a6",
+    "data/cve-catalog.json": "d4d2ecc24a0b010d5e83fea9850130278bea120ef130c99dcbb03eff64008f2d",
+    "data/cwe-catalog.json": "7a8e01c6f5538125b81ddb29f1b658f0fd407682bc2998aafc6b759e3b1afe48",
     "data/d3fend-catalog.json": "a1fc2827ceb344669e148d55197dbf1b0e5b20bcc618e90517639c17d67ee82d",
     "data/dlp-controls.json": "d2406c482dddd30e49203879999dc4b3a7fd4d0494d6a61d86b91ee76415df19",
     "data/exploit-availability.json": "003a400f5ae5b15527589571679ccdb9b3a62e60073627b5fbdeb2a9fe330a7a",
-    "data/framework-control-gaps.json": "45d4b461a733fc31ce5a02b0462d0e23aa5d7469d937d77ad69417eb5293f189",
+    "data/framework-control-gaps.json": "2c37446ad38270ebe51c0eaf340b26f19367c4eae47886d6310f16a4185354bf",
     "data/global-frameworks.json": "9ba563a85f7f8d6c3c957de64945e20925a89d0ed6ea6fc561cf093811acf558",
     "data/rfc-references.json": "e253a548c8a829d178d5aea601e268724b85c936ccbfa51c2e5d80c5f8efe2b0",
     "data/zeroday-lessons.json": "27d46a0e09a3edbe97dfbb070c3991348567cf93c86a3e94c767c5ad2dfb653e",
-    "skills/kernel-lpe-triage/skill.md": "8e94bfd38d6db47342fbbe95a0c8df8f7c38743982c13e9de6a1c59cd3783d33",
-    "skills/ai-attack-surface/skill.md": "853ea46b500fa60b5f5db1137629f8b64447b5df2c8346c15c6cbd1e59285532",
-    "skills/mcp-agent-trust/skill.md": "b09a33e71a0cc13ec70e7e750ac4b91887b657d293d92c3cdb49a4e094adcfea",
-    "skills/framework-gap-analysis/skill.md": "4994b47c2618d24e5557f2a23de21960f6f12a6d66d1b8780b4de6d9d3735dfe",
-    "skills/compliance-theater/skill.md": "5fa6207256d002c42a28a90d15b9a9ef0503ae7dba9b55b4098e2e52607377f4",
-    "skills/exploit-scoring/skill.md": "91bab353257383bd21e49005c1c8188b98b46642184dea29729ad45fc732d698",
-    "skills/rag-pipeline-security/skill.md": "96d26b90539bda836032c2fc2935e0d5736f40bcd286165979c0ed34f47f3ce6",
-    "skills/ai-c2-detection/skill.md": "4f986ac65d4cba36ef9173d204acaf81646a9f7c42623ebba0973ea0108133c0",
-    "skills/policy-exception-gen/skill.md": "78a8623700eab1801387e4792611529089b6248ff3faef15d70c0093609ec323",
-    "skills/threat-model-currency/skill.md": "9fcddaa06334d104cb160bace92c92cdcb6b2881f579b82970e42fbd3d213a05",
-    "skills/global-grc/skill.md": "e0487de49679172347653d8c191d1f269193de6f444f6b0c6396d326e45bd72e",
+    "skills/kernel-lpe-triage/skill.md": "ae4a0af924d0078ffc6cd051a3ef9fce75a6a3f9c0c15d1c07900ae5faf80502",
+    "skills/ai-attack-surface/skill.md": "dcca7d92a1ab4d1e4c46356b614a138b1c1f79b65a6a290eccf2095d8d443993",
+    "skills/mcp-agent-trust/skill.md": "6821f6d38f6e23bbed953f8f86a279597b0b95a2d0548b5383e851bca7442531",
+    "skills/framework-gap-analysis/skill.md": "3b139eaefbedd36b2379cfe22dceef71e97d0e34404b0009b7afbfe0a8dc39e6",
+    "skills/compliance-theater/skill.md": "a1387c523f7aa2481a199f6288e0152b94aa5a6644600eb39dbb3ea9ee9af6bd",
+    "skills/exploit-scoring/skill.md": "fba9e27722d361cc6ed5992d9aaeaa397598b417fc5a0d6fe0bee2993942e7e8",
+    "skills/rag-pipeline-security/skill.md": "ff07e48918090247aef71def4150b0df372a24bcdaa34eb6e11d246b9e71e1ee",
+    "skills/ai-c2-detection/skill.md": "3da9f549f5c62e6163cddd70c8edccbef7be622d5a45fa89c90c6550e68c6b2e",
+    "skills/policy-exception-gen/skill.md": "a7d886f7fa99a150b040f158b09045ba45e107439315389aea785311b0013395",
+    "skills/threat-model-currency/skill.md": "ecc6441cb47ef2bc24547e47be018098228c956a41d61ddb50de7e7b37114a37",
+    "skills/global-grc/skill.md": "1dca534cce7612c1d26a7b1bfd088a811081555ecfa25b1f68cff2ca2ba28c98",
     "skills/zeroday-gap-learn/skill.md": "59a0d7cd85b923b3f5633bdc15c1a88eef7dea6332480d93b0bb0ae93a4cd0fe",
-    "skills/pqc-first/skill.md": "a5eb776e1ea3bb422a4c18a3bdf39ad2ec1651b3c25e65c89428ba319141b275",
-    "skills/skill-update-loop/skill.md": "0d5b08f71e4853a634344eaf260da90bd7a29d4df48200ee75be878dce3467e2",
-    "skills/security-maturity-tiers/skill.md": "2027161ab0a3382ba506cca7be1966e11367bbdf861de316facd54e25fef5761",
-    "skills/researcher/skill.md": "51d03d9eaea52d2bbbdd67709035db494d44819ce58931ca025cab3025c9fad7",
-    "skills/attack-surface-pentest/skill.md": "9cb02d9d428ef674ba8af8c935f86ddca197f0ba1f7d216d76ce1b268ae4bb6a",
-    "skills/fuzz-testing-strategy/skill.md": "eaab866236c8cb8a6c8ddc5e65d786ee6d598682de6014ed4e83c6cd163a2128",
-    "skills/dlp-gap-analysis/skill.md": "ee9fd4928d96b2e9957d8db9dec90c844443fbcf2a292e69040bfa47c78a4f4b",
-    "skills/supply-chain-integrity/skill.md": "497f1ef20f3e506ce1bec5ec65730ffc58300054ff3e8e563e0862838164491a",
+    "skills/pqc-first/skill.md": "a7131b65d0ceee47887b16679ee4e4b065d32d8751fe59921762388703662913",
+    "skills/skill-update-loop/skill.md": "cf2b996cb18a5146614c06e3a50f4734a07d02b5be36bbdf492583f9cdcfed4d",
+    "skills/security-maturity-tiers/skill.md": "ed962937c45f3d95f325f231b787d272fe45c4cb91d4c5a2d982493d722c2acf",
+    "skills/researcher/skill.md": "b47daaa26fdac07aa23e7becaa18487c5302e65c654f99fecab3689f23ec1bd2",
+    "skills/attack-surface-pentest/skill.md": "0d301beb9fb8e247ec80256a7e647804b5f9a41c7156e5724555ca9f93ccb986",
+    "skills/fuzz-testing-strategy/skill.md": "51acb746cd63366ca62567588c700a9eb3f37c43250bd9ae4e1477ccb71c5b6d",
+    "skills/dlp-gap-analysis/skill.md": "1c4e1d7da2421b82f202eaf2c9e21876af34ab5c76ce1359166842ee473f02dd",
+    "skills/supply-chain-integrity/skill.md": "ad69b72f5c5df095f8618b977fbc8f0fbff396eebd4a8448b44c3f93309f63f9",
     "skills/defensive-countermeasure-mapping/skill.md": "e62c71ba3be2b4d0f7dfa529fec007cba6bee3013f76b93756e3e6310f2d22ab",
-    "skills/identity-assurance/skill.md": "6fd734d5cf8eed031537c9ccb1ad11c09ec4e88d31c45d86046a2154a6770990",
-    "skills/ot-ics-security/skill.md": "258003e15bd4e939a1ec1f71c22ffa4dccd6d3194411281f4685abfe840d7cd3",
+    "skills/identity-assurance/skill.md": "4ee7096fd82997c66b0f9e825ea3c04c3aa84768b74e6f668c1a9104104138cf",
+    "skills/ot-ics-security/skill.md": "9ece7b1fb7f24e37dbdd8618b94b2a4434e182e3426e15f17e26464c0a1fdfd1",
     "skills/coordinated-vuln-disclosure/skill.md": "0e875953bb8a38a89c8ec5d2a9ef967b12e9a9f166dc9356723f10304fd0535e",
-    "skills/threat-modeling-methodology/skill.md": "bb34933a1eee2cd1da98da5a5dada2c7fc7ebb0bddf5afb39e1f6ee26064d151",
-    "skills/webapp-security/skill.md": "6fdae41856963df0f8655fb52df7cd26b6b47031f55dfe897003ed9647a73ab5",
-    "skills/ai-risk-management/skill.md": "10d31ca594449e1fef4c34ea45448ab30a6ffdc2fe1faf4ccaf0a1dd05d67774",
-    "skills/sector-healthcare/skill.md": "217066e5961fbc3fcd1b5e3df42f299d7aeaf3b5f25e913152836b77f211f96d",
-    "skills/sector-financial/skill.md": "77f6355eb7672f2157bf3d18bfe1d6042efe302468ebddd48ffc385655bf4d10",
-    "skills/sector-federal-government/skill.md": "981bd7aea1b3ba7875d1222f077126d4eb9462bed46e3e6f5cfed5f94378b152",
-    "skills/sector-energy/skill.md": "44b635b9c52bb10ee2f8d786400926e018c7581387324f59ee3a815de4929c9c",
-    "skills/sector-telecom/skill.md": "3489410b0905cbf6b392ea7f7cde35ccd4b03de0d22d2d1b0c671e46d70962c9",
-    "skills/api-security/skill.md": "75dcb1b9395de2be4ca60e53f900692721b7ef66ded3e510a20d17f35daf982d",
-    "skills/cloud-security/skill.md": "56f0d5d6cf182d347e84baa95a04c39be51e82da3360dac48fcf5d8c4e56a9c3",
-    "skills/container-runtime-security/skill.md": "7e0806b9e13db120f9b65d5f48b33db9f1026c4c2d719838ef0f0c8778ec4365",
-    "skills/mlops-security/skill.md": "a7bf448527e15bb0c6936217c69f3e18cc823f534e0c7f749966cde32039c63e",
-    "skills/incident-response-playbook/skill.md": "0695ee43881527459f657a90276748922347f16dd494ae2b98e2a9396c570a44",
-    "skills/ransomware-response/skill.md": "15de039c5679215b7ceb9a55494f614b06fe618aa0f69ce8aff004dc9a841fa4",
-    "skills/email-security-anti-phishing/skill.md": "b5a7693b3ddbd6cd83303d092bc5e324db431245d25c4945d9f65fcffa1995e7",
-    "skills/age-gates-child-safety/skill.md": "66c7f6537077e4e949214ebf3864cb75de0f50f80fa25d25365c6d7fe485dc7a",
-    "skills/cloud-iam-incident/skill.md": "f124d1170cffb490ddd093ad6fc9876591515643fb2f51379023a08548c7290d",
-    "skills/idp-incident-response/skill.md": "3d75d7a0fc5b9a3c584ac5c6510f8b6bd63b7b780488541eb193250ae795b4e2"
+    "skills/threat-modeling-methodology/skill.md": "ac623f61585de66c9ef5ed63e9c6059faef77e525abc672ac6d435c616a7268f",
+    "skills/webapp-security/skill.md": "fdb07324b69a3a724e3eaba17bf687d72d4bd9d5c4f440be816bc9b08b8aef04",
+    "skills/ai-risk-management/skill.md": "67e62791f60231f2ff53408922fa7137a9060de72097769c630f838a1c227c45",
+    "skills/sector-healthcare/skill.md": "a18e11d25524cdbf40df3798f4c2aa3cb51a4db1b088242ea53fa2885e86b64c",
+    "skills/sector-financial/skill.md": "023b5440d614e6b83ba7294219bcac3cdbffd28fdfdd5f0ec23abbeea71b8230",
+    "skills/sector-federal-government/skill.md": "c63cf1c7c98e920f968cfe60f14e718ea71b120c1b01616af22f64a796963bbe",
+    "skills/sector-energy/skill.md": "643fd951359c2602d9b029a244fe66c1e23f726e711141a06c09cc760a479534",
+    "skills/sector-telecom/skill.md": "862f9482af88e5409e011a6981a5d719863deeb646e41cd4df63e5d6597c50b1",
+    "skills/api-security/skill.md": "2bdfa3dbe534efa3df245e0da37998ad7ab2da4a3171d5000d3346513c10bceb",
+    "skills/cloud-security/skill.md": "c9fad9ed3663cf2faec74ad8f06d62eb86e6636f79933560d8c8d50e0e82d1da",
+    "skills/container-runtime-security/skill.md": "605a8e8eb1af09835b967ec7179456015ec116c6b9051af3a8d225866cc2f7af",
+    "skills/mlops-security/skill.md": "ca3fd922b43fc57aeb5e65c2d5a2823e6bc438167d6afa3a767cee83e4af1f96",
+    "skills/incident-response-playbook/skill.md": "2017515d899c1b2bcb878bc6731e4059623ac52345b2cebbd92204583657bf60",
+    "skills/ransomware-response/skill.md": "2e4fc488f86ed1ba7791ab0e7021160d8ca5ad33a02cdf92a5b916c8afecaa54",
+    "skills/email-security-anti-phishing/skill.md": "e4e9e5a820c0ed3fde9483282e7a0ecaf79284cd2e9923ce66f2b0fb1fc44626",
+    "skills/age-gates-child-safety/skill.md": "66e7773d29c179ab62f409007c05e05993e04a19273225a1e520f2481fd9a90d",
+    "skills/cloud-iam-incident/skill.md": "6494ee3856edeb212e65fe5cdb208357c1a832eb8ac374b26055586bfc71f629",
+    "skills/idp-incident-response/skill.md": "e67a2576e7f1c3bf89f499f5c977bc470ef29e8b3e3e45f4cb5bd45a82674282"
   },
   "skill_count": 42,
   "catalog_count": 11,
@@ -78,7 +78,7 @@
     "handoff_dag_nodes": 42,
     "summary_cards": 42,
     "section_offsets_skills": 42,
-    "token_budget_total_approx": 397775,
+    "token_budget_total_approx": 402643,
     "recipes": 8,
     "jurisdiction_clocks": 29,
     "did_ladders": 8,