ln-625-dependencies-auditor（ln 625 依赖审计员）

Paths: File paths (shared/, references/, ../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root. If shared/ is missing, fetch files via WebFetch from https://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}.

Dependencies & Reuse Auditor (L3 Worker)

Type: L3 Worker

Specialized worker auditing dependency management, code reuse, and security vulnerabilities.

Purpose & Scope

• Worker in ln-760 security-setup pipeline (vulnerabilities_only mode)
• Audit dependencies and reuse (Categories 7+8: Medium Priority)
• Check outdated packages, unused deps, wheel reinvention, CVE vulnerabilities
• Calculate compliance score (X/10)

Parameters

Inputs

MANDATORY READ: Load shared/references/audit_worker_core_contract.md. MANDATORY READ: Load shared/references/mcp_tool_preferences.md and shared/references/mcp_integration_patterns.md

Receives contextStore with tech stack, package manifest paths, codebase root, output_dir.

From ln-620 (codebase-auditor): mode=full (default) From ln-760 (security-setup): mode=vulnerabilities_only

Use hex-graph first when dependency references or code reuse evidence materially improve the audit. Use hex-line first for local code reads when available. If MCP is unavailable, unsupported, or not indexed, continue with built-in Read/Grep/Glob/Bash and state the fallback in the report.

Workflow

MANDATORY READ: Load shared/references/two_layer_detection.md for detection methodology.

1. Parse context + mode parameter + output_dir
2. Run dependency checks (Layer 1: audit tools, based on mode)
3. Analyze context per candidate (Layer 2):

- Available Features: read usage -- is lodash used for 1 function (easy replace) or deeply integrated (hard)? - Custom Implementations: read code -- truly reimplementing a library, or domain-specific logic? - Vulnerability: read code -- is the vulnerable API actually called in this project?

1. Collect findings
2. Calculate score
3. Write Report: Build full markdown report in memory per shared/templates/audit_worker_report_template.md, write to {output_dir}/ln-625--global.md in single Write call
4. Return Summary: Return minimal summary to coordinator

Audit Rules (5 Checks)

1. Outdated Packages

Mode: full only

Detection:

• Run npm outdated --json (Node.js)
• Run pip list --outdated --format=json (Python)
• Run cargo outdated --format=json (Rust)

Severity:

• HIGH: Major version behind (security risk)
• MEDIUM: Minor version behind
• LOW: Patch version behind

Recommendation: Update to latest version, test for breaking changes

Effort: S-M (update version, run tests)

2. Unused Dependencies

Mode: full only

Detection:

• Parse package.json/requirements.txt
• Grep codebase for import/require statements
• Find dependencies never imported

Severity:

• MEDIUM: Unused production dependency (bloats bundle)
• LOW: Unused dev dependency

Recommendation: Remove from package manifest

Effort: S (delete line, test)

3. Available Features Not Used

Mode: full only

Detection:

• Check for axios when native fetch available (Node 18+)
• Check for lodash when Array methods sufficient
• Check for moment when Date.toLocaleString sufficient

Severity:

• MEDIUM: Unnecessary dependency (increases bundle size)

Recommendation: Use native alternative

Effort: M (refactor code to use native API)

4. Custom Implementations

Mode: full only

Detection:

• Grep for custom sorting algorithms
• Check for hand-rolled validation (vs validator.js)
• Find custom date parsing (vs date-fns/dayjs)

Severity:

• HIGH: Custom crypto (security risk)
• MEDIUM: Custom utilities with well-tested alternatives

Recommendation: Replace with established library

Effort: M (integrate library, replace calls)

5. Vulnerability Scan (CVE/CVSS)

Mode: full AND vulnerabilities_only

Detection:

• Detect ecosystems: npm, NuGet, pip, Go, Bundler, Cargo, Composer
• Run audit commands per references/vulnerability_commands.md
• Parse results with CVSS mapping per shared/references/cvss_severity_mapping.md

Severity:

• CRITICAL: CVSS 9.0-10.0 (immediate fix required)
• HIGH: CVSS 7.0-8.9 (fix within 48h)
• MEDIUM: CVSS 4.0-6.9 (fix within 1 week)
• LOW: CVSS 0.1-3.9 (fix when convenient)

Fix Classification:

• Patch update (x.x.Y) -> safe auto-fix
• Minor update (x.Y.0) -> usually safe
• Major update (Y.0.0) -> manual review required
• No fix available -> document and monitor

Recommendation: Update to fixed version, verify lock file integrity

Effort: S-L (depends on breaking changes)

Scoring Algorithm

MANDATORY READ: Load shared/references/audit_worker_core_contract.md and shared/references/audit_scoring.md.

Note: When mode=vulnerabilities_only, score based only on vulnerability findings.

Output Format

MANDATORY READ: Load shared/references/audit_worker_core_contract.md and shared/templates/audit_worker_report_template.md.

Write JSON summary per shared/references/audit_summary_contract.md. In managed mode the caller passes both runId and summaryArtifactPath; in standalone mode the worker generates its own run-scoped artifact path per shared contract.

Write report to {output_dir}/ln-625--global.md with category: "Dependencies & Reuse" and checks: outdated_packages, unused_deps, available_natives, custom_implementations, vulnerability_scan.

Return summary per shared/references/audit_summary_contract.md.

When summaryArtifactPath is absent, write the standalone runtime summary under .hex-skills/runtime-artifacts/runs/{run_id}/evaluation-worker/{worker}--{identifier}.json and optionally echo the same summary in structured output.

Report written: .hex-skills/runtime-artifacts/runs/{run_id}/audit-report/ln-625--global.md
Score: X.X/10 | Issues: N (C:N H:N M:N L:N)

Reference Files

Critical Rules

MANDATORY READ: Load shared/references/audit_worker_core_contract.md.

• Do not auto-fix: Report only, never modify package manifests or lock files
• Mode-aware execution: In vulnerabilities_only mode, skip checks 1-4 entirely
• Effort realism: S = <1h, M = 1-4h, L = >4h
• CVSS-based severity: Map vulnerability severity strictly via shared/references/cvss_severity_mapping.md
• Exclusions: Skip devDependencies for vulnerability severity escalation, skip vendored/bundled deps

Definition of Done

MANDATORY READ: Load shared/references/audit_worker_core_contract.md.

• contextStore parsed (including mode parameter and output_dir)
• All applicable checks completed (5 for full, 1 for vulnerabilities_only)
• Findings collected with severity, location, effort, fix_type, recommendation
• Score calculated per shared/references/audit_scoring.md
• Report written to {output_dir}/ln-625--global.md (atomic single Write call)
• Summary written per contract

Version: 4.0.0 Last Updated: 2026-02-05