ln-760-security-setup（ln 760 安全设置）

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}.

Security Setup Coordinator

Type: L2 Domain Coordinator Category: 7XX Bootstrap

Purpose & Scope

• Coordinate secret scanning (ln-761) and dependency vulnerability audit (ln-625)
• Aggregate findings from both workers into unified report
• Generate security infrastructure: SECURITY.md, pre-commit hooks, CI workflow
• Provide overall security score and risk assessment

When to Use

• During project bootstrap (invoked by ln-700-project-bootstrap)
• Manual security audit request
• CI/CD pipeline initialization

Workflow

Phase 1: Pre-flight Check

Step 1: Detect Project Type

• Identify primary ecosystem(s): Node.js,.NET, Python, Go, etc.
• Check for existing security configs (.gitleaks.toml, SECURITY.md)

Step 2: Check Tool Availability

• Verify gitleaks/trufflehog available for secret scanning
• Verify ecosystem-specific audit tools available
• Log warnings for missing tools (do not fail)

Step 3: Load Existing Configs

• If .gitleaks.toml exists: note for preservation
• If SECURITY.md exists: note for update (not overwrite)
• If .pre-commit-config.yaml exists: check for gitleaks hook

Phase 2: Delegate Scans

Step 1: Invoke ln-761 Secret Scanner

• Delegate via Agent tool
• Receive: findings list, severity summary, remediation guidance

Step 2: Invoke ln-625 Dependencies Auditor (mode=vulnerabilities_only)

• Delegate via Agent tool (can run parallel with Step 1)
• Pass parameter: mode=vulnerabilities_only
• Receive: vulnerability list, CVSS scores, fix recommendations

Phase 3: Aggregate Reports

Step 1: Combine Findings

• Merge findings from both workers
• Group by severity (Critical first)
• Calculate overall security score

Step 2: Risk Assessment

• Critical findings: flag for immediate attention
• High findings: recommend fix within 48h
• Medium/Low: add to backlog

Step 3: Build Summary

• Files scanned count
• Secrets found (by severity)
• Vulnerabilities found (by severity)
• Overall pass/warn/fail status

Phase 4: Generate Outputs

Step 1: Create/Update SECURITY.md

• Use template from references/security_md_template.md
• If exists: update, preserve custom sections
• If new: generate with placeholders

Step 2: Configure Pre-commit Hooks

• If .pre-commit-config.yaml missing: create from template
• If exists without gitleaks: recommend adding
• Template: references/precommit_config_template.yaml

Step 3: Generate CI Workflow

• If .github/workflows/security.yml missing: create from template
• Template: references/ci_workflow_template.yaml
• Include ecosystem-specific audit jobs

Step 4: Update.gitignore

• Ensure secret-related patterns present:

- .env, .env.*, !.env.example - *.pem, *.key

• Preserve existing entries

Worker Invocation (MANDATORY)

CRITICAL: All delegations use Agent tool with subagent_type: "general-purpose" for context isolation.

Prompt template:

Agent(description: "Secret scanning via ln-761",
     prompt: "Execute security scanner.

Step 1: Invoke worker:
  Skill(skill: \"ln-761-secret-scanner\")

CONTEXT:
Project: {projectPath}",
     subagent_type: "general-purpose")

Agent(description: "Dependency vulnerability scan via ln-625",
     prompt: "Execute vulnerability scanner.

Step 1: Invoke worker:
  Skill(skill: \"ln-625-dependencies-auditor\")

CONTEXT:
Project: {projectPath}
Mode: vulnerabilities_only (only CVE scan, skip outdated/unused checks)",
     subagent_type: "general-purpose")

Pattern: Both workers can execute in parallel via Agent tool, then aggregate results.

Anti-Patterns:

• ❌ Direct Skill tool invocation without Agent wrapper
• ❌ Any execution bypassing subagent context isolation
• ❌ Calling ln-625 without mode parameter (would run full audit)

Definition of Done

• Both workers (ln-761, ln-625) invoked and completed
• Findings aggregated with severity classification
• SECURITY.md created/updated
• Pre-commit hook configured (or recommendation logged)
• CI workflow generated (or recommendation logged)
• .gitignore updated with secret patterns
• Summary report returned to parent orchestrator

TodoWrite format (mandatory)

- Phase 1: Invoke secret scanner ln-761 (pending)
- Phase 2: Invoke dependency auditor ln-625 (pending)
- Phase 3: Aggregate findings (pending)
- Phase 4: Generate security artifacts (pending)
- Phase 5: Return summary (pending)

Meta-Analysis

MANDATORY READ: Load shared/references/meta_analysis_protocol.md

Skill type: planning-coordinator. Run after all phases complete. Output to chat using the protocol format.

Reference Files

Critical Rules

• Always pass mode=vulnerabilities_only to ln-625 — full audit mode is not appropriate for bootstrap context
• Preserve existing configs — if .gitleaks.toml, SECURITY.md, or .pre-commit-config.yaml exist, update rather than overwrite
• Use Agent tool with subagent_type: "general-purpose" for all worker delegations (context isolation)
• Never fail on missing tools — log warnings for unavailable scanners, continue with available ones
• Critical findings block completion — flag for immediate attention before returning to parent

Version: 3.0.0 Last Updated: 2026-02-05