@softspark/ai-toolkit 4.25.0 → 4.25.1

package/CHANGELOG.md

@@ -7,6 +7,30 @@ Versioning follows [Semantic Versioning](https://semver.org/).
 
 ---
 
+## v4.25.1 — npm advisories count again (2026-08-19)
+
+### Fixed
+
+- **`cve-scan` reported zero findings for a vulnerable npm project.** npm audit
+  can return the legacy `advisories` object as well as the current
+  `vulnerabilities` object. The scanner parsed only the latter, so five real
+  lodash advisories remained trapped in `raw_output` while `total_findings` was
+  zero and the process exited successfully. Both formats are now normalized,
+  and HIGH advisories restore the documented non-zero exit code.
+
+### Changed
+
+- **The published-package scanner smoke uses real per-scanner fixtures and
+  flags.** It now includes healthcare patterns and a deliberately vulnerable
+  locked npm dependency, invokes `cve-scan` with `--json`, and treats raw npm
+  advisories with no normalized findings as a parser failure.
+
+### Tests
+
+- Added a public-CLI regression test with a stubbed legacy npm audit response;
+  it asserts normalized severity, installed version, CVE, fixed range, finding
+  count, and exit status.
+
 ## v4.25.0 — editor-native integrations catch up (2026-08-19)
 
 ### Added

package/README.md

@@ -6,20 +6,18 @@
 [![License: Apache 2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)
 [![Skills](https://img.shields.io/badge/skills-109-brightgreen)](app/skills/)
 [![Agents](https://img.shields.io/badge/agents-44-blue)](app/agents/)
-[![Tests](https://img.shields.io/badge/tests-1636%20passing-success)](tests/)
+[![Tests](https://img.shields.io/badge/tests-1637%20passing-success)](tests/)
 
-## What's New in v4.25.0
+## What's New in v4.25.1
 
-**v4.25.0** brings the editor integrations onto their current native surfaces:
+**v4.25.1** fixes the CVE scanner path exercised by the published-package smoke test:
 
-- Codex gets its exact 11-event hook contract and a deterministic native plugin
-  exporter with offline verification.
-- Google Antigravity gets native hooks, agents, plugin packaging, MCP mapping,
-  and local/global install profiles.
-- Gemini gains native agents, OpenCode gains native skill directories, and Cline
-  gains native rules plus its exact 8-event hook lifecycle.
-- Claude Code validation now matches the current prompt, agent, HTTP hook, and
-  plugin contracts, with transactional symlink protections across new writers.
+- Legacy npm audit `advisories` are normalized alongside the modern
+  `vulnerabilities` response.
+- HIGH advisories now produce findings and the documented non-zero exit code
+  instead of a false clean result.
+- The post-release scanner fixture now exercises accessibility, SEO, HIPAA, and
+  a deliberately vulnerable npm dependency with their real CLI flags.
 
 See [CHANGELOG.md](CHANGELOG.md) for full history.
 
@@ -185,7 +183,7 @@ ai-toolkit/
 │   └── ARCHITECTURE.md  # Full system design
 ├── kb/                  # Reference docs, procedures, plans
 ├── scripts/             # Validation, install, evaluation scripts
-├── tests/               # Bats and Python test suite (1636 tests)
+├── tests/               # Bats and Python test suite (1637 tests)
 └── CHANGELOG.md
 ```
 

package/app/.claude-plugin/plugin.json

@@ -3,7 +3,7 @@
   "name": "ai-toolkit",
   "displayName": "AI Toolkit",
   "description": "Professional-grade engineering skills, agents, rules, and lifecycle guardrails for Claude Code, Claude Chat, and Cowork.",
-  "version": "4.25.0",
+  "version": "4.25.1",
   "author": {
     "name": "SoftSpark",
     "url": "https://github.com/softspark"

package/app/skills/cve-scan/scripts/cve_scan.py

@@ -9,7 +9,6 @@ Stdlib only. No external dependencies.
 """
 
 import json
-import os
 import shutil
 import subprocess
 import sys
@@ -141,7 +140,7 @@ def run_audit(eco: dict, root: Path, fix: bool = False) -> dict:
     except FileNotFoundError:
         result["error"] = f"Tool '{eco['tool']}' not found in PATH"
     except subprocess.TimeoutExpired:
-        result["error"] = f"Audit command timed out after 120s"
+        result["error"] = "Audit command timed out after 120s"
 
     return result
 
@@ -168,6 +167,21 @@ def parse_npm_audit(raw: str) -> list[dict]:
                     "url": via.get("url", ""),
                     "fixed_in": info.get("fixAvailable", {}).get("version", "unknown") if isinstance(info.get("fixAvailable"), dict) else "unknown",
                 })
+
+    for advisory in data.get("advisories", {}).values():
+        installed_versions = sorted({
+            finding.get("version", "unknown")
+            for finding in advisory.get("findings", [])
+        })
+        findings.append({
+            "severity": advisory.get("severity", "unknown").upper(),
+            "package": advisory.get("module_name", "unknown"),
+            "installed": ", ".join(installed_versions) or "unknown",
+            "cve": advisory.get("cves", []) or [advisory.get("github_advisory_id", "N/A")],
+            "title": advisory.get("title", "Unknown"),
+            "url": advisory.get("url", ""),
+            "fixed_in": advisory.get("patched_versions", "unknown"),
+        })
     return findings
 
 

package/kb/procedures/post-release-testing-sop.md

@@ -3,9 +3,9 @@ title: "SOP: Post-Release Testing"
 category: procedures
 service: ai-toolkit
 tags: [sop, post-release, smoke-test, npm, sandbox, plugin-pack, provenance, isolation]
-version: "1.1.0"
+version: "1.2.0"
 created: "2026-07-26"
-last_updated: "2026-08-06"
+last_updated: "2026-08-19"
 description: "Smoke-test a published @softspark/ai-toolkit release from npm in an isolated HOME and npm prefix, without touching the maintainer's real install. Covers provenance, CLI, doctor, per-skill script resolution, scanner wiring, and the full plugin-pack lifecycle including the degraded-install path. Written for v4.18.0 and not run; v4.18.0 shipped a pack that broke every command it touched. First actually run on v4.22.0, which added Phases 4b and 4c after that release fixed four skills whose documented script path had never resolved and two that shipped a scanner nothing invoked."
 ---
 
@@ -159,21 +159,30 @@ cat > "$FX/index.html" <<'EOF'
 <!DOCTYPE html><html><head><title>t</title></head>
 <body><h1>A</h1><h3>skipped h2</h3><img src="x.png"><input type="text"><div onclick="go()">click</div></body></html>
 EOF
-printf '{"name":"fx","dependencies":{"react":"18"}}\n' > "$FX/package.json"
-
-for s in a11y-validate seo-validate hipaa-validate cve-scan; do
-  D="$HOME/.claude/skills/$s"
-  script=$(ls "$D"/scripts/*.py 2>/dev/null | head -1)
-  [ -n "$script" ] || continue
-  echo "=== $s ==="
-  CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$script" "$FX" --output json </dev/null 2>&1 | head -3
-done
+cat > "$FX/patient.py" <<'EOF'
+patient = {"ssn": "123-45-6789", "diagnosis": "fixture"}
+endpoint = "http://example.com/patient"
+encrypt = False
+EOF
+printf '{"name":"fx","dependencies":{"lodash":"4.17.20","react":"18"}}\n' > "$FX/package.json"
+npm install --package-lock-only --ignore-scripts --prefix "$FX"
+
+D="$HOME/.claude/skills/a11y-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/a11y-scanner.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/seo-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/seo-scanner.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/hipaa-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/hipaa_scan.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/cve-scan"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/cve_scan.py" "$FX" --json </dev/null
 ```
 
 **Verify:**
 - [ ] The scanner returns findings, not an empty set — the fixture has real defects
 - [ ] Findings span more than one category, proving the whole check set ran
 - [ ] A scanner that exits non-zero on findings is doing its job, not failing
+- [ ] CVE output has `total_findings > 0`; raw npm advisories with an empty
+      normalized `findings` array are a parser failure, not a clean scan
 
 `a11y-validate` and `seo-validate` shipped working scanners that **no step in
 either skill invoked** for their entire life before v4.22.0. The model was told to

package/kb/reference/enterprise-config-guide.md

@@ -11,7 +11,7 @@ tags:
 doc_type: reference
 created: "2026-04-11"
 last_updated: "2026-07-26"
-description: "Comprehensive guide for setting up and using ai-toolkit configuration inheritance. Covers base config creation, project setup, enforcement rules, CI integration, and troubleshooting."
+description: "Configuration inheritance for ai-toolkit projects extends base settings from npm packages, Git URLs, or local paths. Defines merge semantics, enforcement rules, and CLI commands for validating and diffing configurations."
 ---
 
 # Enterprise Config Inheritance Guide

package/llms-full.txt

@@ -7041,9 +7041,9 @@ title: "SOP: Post-Release Testing"
 category: procedures
 service: ai-toolkit
 tags: [sop, post-release, smoke-test, npm, sandbox, plugin-pack, provenance, isolation]
-version: "1.1.0"
+version: "1.2.0"
 created: "2026-07-26"
-last_updated: "2026-08-06"
+last_updated: "2026-08-19"
 description: "Smoke-test a published @softspark/ai-toolkit release from npm in an isolated HOME and npm prefix, without touching the maintainer's real install. Covers provenance, CLI, doctor, per-skill script resolution, scanner wiring, and the full plugin-pack lifecycle including the degraded-install path. Written for v4.18.0 and not run; v4.18.0 shipped a pack that broke every command it touched. First actually run on v4.22.0, which added Phases 4b and 4c after that release fixed four skills whose documented script path had never resolved and two that shipped a scanner nothing invoked."
 ---
 
@@ -7197,21 +7197,30 @@ cat > "$FX/index.html" <<'EOF'
 <!DOCTYPE html><html><head><title>t</title></head>
 <body><h1>A</h1><h3>skipped h2</h3><img src="x.png"><input type="text"><div onclick="go()">click</div></body></html>
 EOF
-printf '{"name":"fx","dependencies":{"react":"18"}}\n' > "$FX/package.json"
-
-for s in a11y-validate seo-validate hipaa-validate cve-scan; do
-  D="$HOME/.claude/skills/$s"
-  script=$(ls "$D"/scripts/*.py 2>/dev/null | head -1)
-  [ -n "$script" ] || continue
-  echo "=== $s ==="
-  CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$script" "$FX" --output json </dev/null 2>&1 | head -3
-done
+cat > "$FX/patient.py" <<'EOF'
+patient = {"ssn": "123-45-6789", "diagnosis": "fixture"}
+endpoint = "http://example.com/patient"
+encrypt = False
+EOF
+printf '{"name":"fx","dependencies":{"lodash":"4.17.20","react":"18"}}\n' > "$FX/package.json"
+npm install --package-lock-only --ignore-scripts --prefix "$FX"
+
+D="$HOME/.claude/skills/a11y-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/a11y-scanner.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/seo-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/seo-scanner.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/hipaa-validate"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/hipaa_scan.py" "$FX" --output json </dev/null
+D="$HOME/.claude/skills/cve-scan"
+CLAUDE_SKILL_DIR="$D" timeout 60 python3 "$D/scripts/cve_scan.py" "$FX" --json </dev/null
 ```
 
 **Verify:**
 - [ ] The scanner returns findings, not an empty set — the fixture has real defects
 - [ ] Findings span more than one category, proving the whole check set ran
 - [ ] A scanner that exits non-zero on findings is doing its job, not failing
+- [ ] CVE output has `total_findings > 0`; raw npm advisories with an empty
+      normalized `findings` array are a parser failure, not a clean scan
 
 `a11y-validate` and `seo-validate` shipped working scanners that **no step in
 either skill invoked** for their entire life before v4.22.0. The model was told to
@@ -11046,7 +11055,7 @@ tags:
 doc_type: reference
 created: "2026-04-11"
 last_updated: "2026-07-26"
-description: "Comprehensive guide for setting up and using ai-toolkit configuration inheritance. Covers base config creation, project setup, enforcement rules, CI integration, and troubleshooting."
+description: "Configuration inheritance for ai-toolkit projects extends base settings from npm packages, Git URLs, or local paths. Defines merge semantics, enforcement rules, and CLI commands for validating and diffing configurations."
 ---
 
 # Enterprise Config Inheritance Guide

package/manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "4.25.0",
+  "version": "4.25.1",
   "components": {
     "agents": {
       "description": "44 specialized agents (orchestrator, backend, frontend, security, devops, etc.)",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@softspark/ai-toolkit",
-  "version": "4.25.0",
+  "version": "4.25.1",
   "description": "AI coding toolkit: 109 skills, 44 agents, 12 developer-tool integrations, recoverable native tool-output filtering, Claude Chat/Cowork export, safety constitution, SARIF audit, and signed npm provenance.",
   "keywords": [
     "claude",