skills/docs-update/SKILL.md
Update documentation by synchronizing with the current codebase state. This skill should be used when users want to refresh stale documentation, sync docs with code changes, or update specific documentation files or categories.
npx skillsauth add mgiovani/cc-arsenal docs-updateInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Cross-Platform AI Agent Skill This skill works with any AI agent platform that supports the skills.sh standard.
Synchronize documentation with the current codebase state. Can update all docs, specific files, or entire categories.
CRITICAL: Documentation updates must reflect REALITY, not assumptions:
Before updating ANY documentation, thoroughly explore the codebase:
For updating multiple documents, use TodoWrite to track progress:
Create todos for each document to update, marking them in_progress as you work
### Phase 3: Parse Arguments
1. Extract update mode from `command arguments`
2. Modes:
- No args or `all` -> Update all relevant docs
- `<doc-name>` -> Update specific doc (e.g., `architecture`)
- `category:<name>` -> Update category (e.g., `category:data`)
3. Default: Update all
### Phase 4: Determine Update Scope and Analyze
**For "all" mode**: Analyze entire project, identify all relevant documentation types, update everything that exists or should exist.
**For specific doc mode**: Validate doc name, find the corresponding file, update only that document.
**For category mode**: Parse category name (`core`, `data`, `infrastructure`, `development`), identify all docs in that category, update all.
### Phase 5: Check Documentation Freshness
```bash
# For each doc, compare with related code changes
!`git log --since="$(git log -1 --format=%ai docs/architecture.md)" --oneline --name-only | head -30`
Identify which docs are outdated and prioritize updates.
### Phase 6: Parallel Updates (Use Parallel Analysis)
For multiple document updates, see [references/update-strategies.md](references/update-strategies.md) for parallel subagent patterns.
#### For Multiple Documents ("all" or "category" mode)
#### For Single Document (Section-Level Parallelization)
Even when updating ONE document, spawn subagents for each major section. See [references/update-strategies.md](references/update-strategies.md) for section-level patterns.
### Phase 7: Update Each Document
- Re-analyze relevant parts of codebase
- **Verify each claim before writing** - Read actual files
- Regenerate diagrams if needed
- Update content while preserving custom sections
- Replace placeholders with current values
- **Remove claims that cannot be verified**
### Phase 8: Post-Update Verification
After updating, verify the updates are accurate:
For each updated document:
Files that are always relevant:
docs/architecture.md - System architecturedocs/onboarding.md - Developer onboardingdocs/adr/ - Architecture Decision RecordsUpdate when: Architecture changes, setup process changes
Files relevant if database detected:
docs/data-model.md - Database schema and ER diagramsUpdate when: Schema changes, models modified
Files relevant if deployment configs detected:
docs/deployment.md - CI/CD and deploymentdocs/security.md - Security architectureUpdate when: Infrastructure changes, security updates
Files relevant for collaborative projects:
docs/contributing.md - Contribution guidelinesdocs/rfc/ - RFC documentsdocs/api-documentation.md - API documentationUpdate when: Process changes, API changes
| Argument | File Path |
|----------------|----------------------------|
| architecture | docs/architecture.md |
| onboarding | docs/onboarding.md |
| data-model | docs/data-model.md |
| deployment | docs/deployment.md |
| security | docs/security.md |
| contributing | docs/contributing.md |
| api-docs | docs/api-documentation.md|
Update all documentation:
docs-update
docs-update all
Update specific document:
docs-update architecture docs-update data-model docs-update onboarding docs-update deployment docs-update security Update entire category:
docs-update category:core
docs-update category:data
docs-update category:infrastructure
docs-update category:development
With context:
docs-update architecture after microservices refactor docs-update data-model after schema migration docs-update category:core for onboarding review
Replace these placeholders during update:
{{PROJECT_NAME}} - Current project name{{DATE}} - Current date{{TECH_STACK}} - Detected technologies{{ER_DIAGRAM}} - Generated ER diagram{{ARCHITECTURE_DIAGRAM}} - Generated architecture diagram{{ENTITIES}} - Extracted entity descriptions{{COMPONENTS}} - Extracted component descriptionsThis skill includes the following Claude Code-specific enhancements:
Before updating ANY documentation, thoroughly explore the codebase:
Use Task tool with Explore agent:
- prompt: "Comprehensively analyze this codebase. Find: 1) All actual source files and their purposes, 2) Real component counts (services, models, APIs), 3) Actual directory structure with content verification, 4) Technologies actually in use (check package files). Return ONLY verified facts with file paths as evidence."
- subagent_type: "Explore"
For updating multiple documents, use TodoWrite to track progress:
Create todos for each document to update, marking them in_progress as you work
$ARGUMENTSall -> Update all relevant docs<doc-name> -> Update specific doc (e.g., architecture)category:<name> -> Update category (e.g., category:data)For "all" mode: Analyze entire project, identify all relevant documentation types, update everything that exists or should exist.
For specific doc mode: Validate doc name, find the corresponding file, update only that document.
For category mode: Parse category name (core, data, infrastructure, development), identify all docs in that category, update all.
# For each doc, compare with related code changes
!`git log --since="$(git log -1 --format=%ai docs/architecture.md)" --oneline --name-only | head -30`
Identify which docs are outdated and prioritize updates.
For multiple document updates, see references/update-strategies.md for parallel subagent patterns.
Use Task tool with multiple parallel agents:
Agent 1 - Architecture Update:
- prompt: "Update docs/architecture.md. Explore codebase, verify claims, remove false info, add missing components."
- subagent_type: "general-purpose"
Agent 2 - Data Model Update:
- prompt: "Update docs/data-model.md. Find actual models, verify ER diagram accuracy."
- subagent_type: "general-purpose"
Agent 3 - Onboarding Update:
- prompt: "Update docs/onboarding.md. Verify setup instructions and commands exist."
- subagent_type: "general-purpose"
Even when updating ONE document, spawn subagents for each major section. See references/update-strategies.md for section-level patterns.
After updating, verify the updates are accurate:
For each updated document:
1. Re-read the document
2. For each major claim, verify against actual code
3. If any claim cannot be verified, remove it
4. Run /docs-check to validate
development
Generate comprehensive test suites with coverage analysis and parallel test writing. Automatically activates when users want to write tests, add test coverage, generate test cases, improve testing, or analyze coverage gaps. Supports pytest, vitest, jest, and all major test frameworks.
development
Multi-agent PR review team orchestration with 7 specialized reviewers for security-sensitive or architectural PRs. Spawns architecture, security, performance, testing, style, docs/UX, and adversary reviewers as a coordinated team. Premium review for critical code changes.
development
Spec-driven team orchestration: adaptive development team scaling from 3 to 11 agents based on complexity.
development
Perform comprehensive security review targeting OWASP Top 10 2025 vulnerabilities for PRs, commits, or entire codebases. This skill should be used when a user wants to audit code security, scan for vulnerabilities, review security posture, or check for OWASP compliance. Analysis only - identifies vulnerabilities without modifying code.