Agent Assurance — Security skill for Claude Code
Declare what your AI agent may do.
How to install Agent Assurance
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open kunko-ai-labs/agent-assurance and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Agent Assurance does
Declare what your AI agent may do. Verify it on every edit, PR and release — MCP configs, Claude Code permissions, tool definitions. Deterministic, OWASP-mapped, signed evidence.
Alternatives in Security
- Anthropic Cybersecurity Skills — 734+ structured cybersecurity skills for AI agents · MITRE ATT&CK mapped · agentskills.io open standard · Work 3.8k ★
- Skill Audit — Audit codebases for quality, consistency, and broken patterns — use for pre-release or tech debt review 2.8k ★
- Trail Of Bits Claude-code-config — Opinionated production defaults from a top security firm: sandboxing, permissions, hooks, skills, MCP server c 1.6k ★
README
Agent Assurance
[](https://github.com/kunko-ai-labs/agent-assurance/actions/workflows/ci.yml) [](https://github.com/kunko-ai-labs/agent-assurance/actions/workflows/assurance.yml) [](LICENSE)
**Declare what your agent may do. Verify it on every edit, every PR, every release.**
Your repo *promises* what an AI agent is allowed to do (`agent-assurance.yaml`: read the CRM, no external send, a human approves actions). Agent Assurance *observes* what the configuration actually grants — MCP servers, Claude Code permissions, tool definitions — and fails the change when the promise is broken, pointing at the file and line that broke it. Then it leaves a signed record of what the agent could do, and when.
No dashboard, no backend, no LLM in the verdict, no network calls, nothing executed. Rules you can read; numbers you can reproduce by hand.
pipx install agent-assurance
agent-assurance scan . # what does this repo let the agent do — and does it match the promise?▶ [22-second launch video](https://github.com/kunko-ai-labs/agent-assurance/releases/download/v0.5.0/brag.mp4)
What it catches
| Situation | What Agent Assurance says |
|---|---|
A PR adds @modelcontextprotocol/server-github to an agent declared read-only |
Promise broken: github.write grants write, not declared (.mcp.json:17); blast radius LOW → HIGH |
.claude/settings.json gets allow: Bash(*) while the manifest says a human approves (L2) |
Promise broken: Bash(*) runs without human approval, but declared autonomy is L2 |
| A server config carries `GITHUB_PERSONAL_ACCES |
Related Skills
GitHub Release Skill
Safe, automated GitHub releases with supply chain security. Prevents dangerous gh release commands, orchestrat
Hermes Repo Scanner
Pre-install security scanner for Hermes Agent skills — 68 patterns across 10 categories, OWASP ASI-mapped
Ke Autonomous
Orchestrate an approved project through product analysis, architecture, implementation, automated testing, QA,
Ke Ship
Prepare a release with evidence, compatibility, security, rollback, and post-release checks; stop before exter
Security Gate
Security verification for G4 — verify allocated controls actually exist and work, triage scanner output, produ
Sentience Governor
Local-first runtime governance for AI agents. Declare intent, catch out-of-scope actions, and audit them in a
Related Agents
Cook Verify
jeff verify stage for operation tasks. Independently verify every deterministic postcondition and acceptance c
Vkplay Builder
Builds, validates, and packages an HTML5 game for VK Play (vkplay.ru) — iframe-embedded with signed auth + pay
Launch Coordinator
Use to prepare release notes and a rollout/rollback checklist once the CEO review has signed off. Invoke for t
