- Home
- Skills
- Agents & Orchestration
- ai-agent-kit
ai Agent Kit
A multi-repo governance kit for deep code auditing, rule enforcement, and structured task implementation.
Free
Works with the AI Agents you already use
ai-agent-kit
Example session with this skill installed
/x-check --severity high
- Read your context and instructions
- Compiled the ai-agent-kit
I have audited the staged and unstaged changes in the backend and frontend repos against the project rules.
/x-check report — High/Critical Severities
Repos scanned: backend, frontend
Files checked: 4
Rules applied: 12 (Filter: high)
Violations found: 2
🔴 Critical (1)
Line 45: S-02 No Hardcoded Secrets
const API_KEY = "sk-live-51MzS...";
Fix: Move all secrets to environment variables or a secure vault. Use process.env.STRIPE_SECRET.
🟡 High (1)
Line 102: A-05 Missing CSRF Protection
@Post('/update-profile')
async updateProfile(@Body() data: UpdateDto) {
Fix: Ensure all state-changing routes have the @UseGuards(CsrfGuard) decorator or equivalent middleware.
Violations by rule
| Rule | Lines |
|---|---|
S-02 No Hardcoded Secrets | backend/src/config.ts:45 |
A-05 Missing CSRF Protection | backend/src/user.controller.ts:102 |
Next steps
- Apply the fixes to the identified lines in
backend. - Run
/x-check --staged-onlyafter fixing to verify compliance. - Call
/x-add-ruleif this audit surfaced a new pattern to block.
Connects securely to your tools. The creator never sees your data.
About this skill
Codebases often suffer from "rule drift" where architectural constraints and security standards are documented in wikis but ignored in PRs. Manual code reviews are slow, inconsistent, and prone to missing non-obvious violations like Unicode homoglyphs or specific AI-generated patterns.
What it does
- Multi-repo Auditing: Performs line-by-line senior-level code reviews across NestJS, Laravel, Next.js, and Flutter projects.
- Rule Enforcement: Checks uncommitted or staged changes against a central
rules.mdfile using regex and semantic analysis. - Automated Implementation: Executes complex coding tasks in three phases: scope-locking, implementation, and automated self-correction.
- Rule Lifecycle Management: Appends new project-specific rules with structured Why/Detect/Fix fields and auto-updates project metadata.
- Change Verification: Generates clickable reports for rule violations, including specific line numbers and required fixes.
How it works
- Audit: Use
/x-auditto map the codebase and generate structured architecture and findings reports. - Implement: Use
/x-implementfor tasks, which triggers a Phase 0 scope contract before writing any code. - Check: Run
/x-checkor/x-check-fileto scan local changes against the project's security and quality rules. - Govern: Use
/x-add-ruleto codify new constraints learned from audits or bugs into the master rule file.
Frameworks & tools
This kit supports major stacks including NestJS, Express, Laravel, Django, Next.js, React, Flutter, and Go. It utilizes standard CLI tools like grep, bash, and git for deep inspection.
Why this beats prompting it yourself
Generic prompts forget project-specific constraints after a few turns. This skill forces every interaction through a persistent, version-controlled rules.md file, ensuring that architectural "laws" are enforced pedantically across entire sessions and multi-repo environments.
Use cases
- Enforce strict authentication patterns across a monorepo.
- Perform a deep security audit of a legacy PHP or Node.js codebase.
- Scaffold features with mandatory Phase 0 planning to prevent scope creep.
- Automatically scan for "AI-tell" vocabulary and non-standard Unicode in PRs.
Known limitations
Audits of files exceeding 2000 lines require chunked reading. Rule detection depends on the quality of the plain-English 'Detect' instructions in the master rules file.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Requires a RULES_PATH file to exist. Large files (>2000 lines) are processed in chunks. Unicode scanning requires specific grep patterns.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
This skill has been used 1 time.
Trust & safety
Security scan
Security scan passed 17 days ago
- Free with an Agensi account
- Software
Git