Skip to content
Back to search

agent-security-playbook

MarketplaceWorks with Claude CodeGitHub

OWASP Secure Agent Playbook — security skills for code analysis and AI/agent security

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/owasp--secure-agent-playbook/badge.svg)](/listings/owasp--secure-agent-playbook)

HTML

<a href="/listings/owasp--secure-agent-playbook"><img src="/api/listings/owasp--secure-agent-playbook/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

**With Claude Code (recommended):**

**Step 1** — Register the plugin marketplace:

/plugin marketplace add OWASP/secure-agent-playbook

**Step 2** — Install a skill set:

/plugin install code-security-skills@agent-security-playbook
/plugin install ai-security-skills@agent-security-playbook

**Step 3** — Use the skills by mentioning the task in conversation:

"Review this code for security issues"
"Scan my dependencies for CVEs"
"Audit this MCP server configuration"
"Test this chatbot for prompt injection"

Claude will automatically activate the relevant skill based on context. See [Skills Catalog](#skills-catalog) for all available skills and [Example Output](#example-output) for what the results look like.

**Organization plugin** — For Claude organization admins installing via [Organization Plugins](https://claude.ai/admin-settings/plugins):

1. Go to the [latest release](https://github.com/OWASP/secure-agent-playbook/releases/latest) 2. Download `secure-agent-playbook.zip` from the release assets 3. Upload the zip at **Organization settings > Plugins > Add plugin**

> **Note:** Do not use GitHub's "Download ZIP" button — it nests files in a subdirectory that the plugin validator rejects. Always use the release asset zip.

**Local development** — To test from a local clone instead of GitHub:

/plugin marketplace add /path/to/agent-security-playbook
/plugin install code-security-skills@agent-security-playbook
/plugin install ai-security-skills@agent-security-playbook

**Without Claude Code:**

Reference plays directly as procedures for any AI agent or manual use: - Point your agent at a play: *"Follow the procedure in `plugins/ai-security-skills/plays/agent-security-audit.md`"* - Or use the plays as checklists for manual security reviews

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 20 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "name": "agent-security-playbook",
  3. "description": "OWASP Secure Agent Playbook — security skills for code analysis and AI/agent security",
  4. "owner": {
  5. "name": "Chris Maenner",
  6. "email": "chris@palmtreeventuresllc.com"
  7. },
  8. "plugins": [
  9. {
  10. "name": "code-security-skills",
  11. "description": "Code and infrastructure security analysis skills — security engineering, code review, SCA audit, secrets scanning, API security review, web application security review, and IaC security review.",
  12. "source": "./plugins/code-security-skills/"
  13. },
  14. {
  15. "name": "ai-security-skills",
  16. "description": "AI and agent security assessment skills — agent security audit, LLM risk assessment, agentic AI risk assessment, MCP server review, prompt injection testing, multi-agent threat modeling, and AISVS verification.",
  17. "source": "./plugins/ai-security-skills/"
  18. }
  19. ]
  20. }
README.md · README · 268 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Secure Agent Playbook
  2. An open-source security playbook for AI agents. Structured, OWASP-grounded procedures that enable agents to perform security engineering tasks — from code review to AI agent security audits.
  3. ## Table of Contents
  4. - [Why Use This?](#why-use-this)
  5. - [What This Is](#what-this-is)
  6. - [Quick Start](#quick-start)
  7. - [Skills Catalog](#skills-catalog)
  8. - [Agents](#agents)
  9. - [Example Output](#example-output)
  10. - [Plays](#plays)
  11. - [Architecture](#architecture)
  12. - [OWASP Foundation](#owasp-foundation)
  13. - [Related Projects](#related-projects)
  14. - [Contributing](#contributing)
  15. ## Why Use This?
  16. Without a playbook, asking an AI agent to "review my code for security" gives you a surface-level checklist. With these plays, the agent follows a structured OWASP-grounded procedure — systematically testing every vulnerability class, producing findings with CWE mappings, OpenCRE cross-references, evidence snippets, and specific remediation code.
  17. - **Consistent methodology** — Every assessment follows a documented procedure, not ad-hoc prompting. Results are reproducible across runs and reviewers.
  18. - **Structured, actionable output** — Findings include severity, CWE, evidence, and remediation steps with code examples. No vague warnings.
  19. - **Cross-standard traceability** — Findings link to CWE, ASVS, WSTG, and NIST 800-53 via [OpenCRE](https://www.opencre.org) for compliance mapping.
  20. - **18 security skills** — From dependency CVE scanning to prompt injection testing to multi-agent threat modeling. Install as a Claude Code plugin or use standalone.
  21. - **Works beyond Claude Code** — Skills are Claude Code plugins; plays are standalone procedures any AI agent can follow.
  22. ## What This Is
  23. This is not a framework or a library. There is no code to import.
  24. Each **play** is a step-by-step security procedure with checklists, decision criteria, and output templates. An AI agent follows the procedure to produce consistent, evidence-based findings. Think of it like a SOC analyst's playbook — but written for AI agents to execute.
  25. ## Quick Start
  26. **With Claude Code (recommended):**
  27. **Step 1** — Register the plugin marketplace:
  28. ```
  29. /plugin marketplace add OWASP/secure-agent-playbook
  30. ```
  31. **Step 2** — Install a skill set:
  32. ```
  33. /plugin install code-security-skills@agent-security-playbook
  34. /plugin install ai-security-skills@agent-security-playbook
  35. ```
  36. **Step 3** — Use the skills by mentioning the task in conversation:
  37. ```
  38. "Review this code for security issues"
  39. "Scan my dependencies for CVEs"
  40. "Audit this MCP server configuration"
  41. "Test this chatbot for prompt injection"
  42. ```
  43. Claude will automatically activate the relevant skill based on context. See [Skills Catalog](#skills-catalog) for all available skills and [Example Output](#example-output) for what the results look like.
  44. **Organization plugin** — For Claude organization admins installing via [Organization Plugins](https://claude.ai/admin-settings/plugins):
  45. 1. Go to the [latest release](https://github.com/OWASP/secure-agent-playbook/releases/latest)
  46. 2. Download `secure-agent-playbook.zip` from the release assets
  47. 3. Upload the zip at **Organization settings > Plugins > Add plugin**
  48. > **Note:** Do not use GitHub's "Download ZIP" button — it nests files in a subdirectory that the plugin validator rejects. Always use the release asset zip.
  49. **Organization skills** — For Claude organization admins uploading individual skills at [Organization settings > Skills](https://claude.ai/admin-settings/skills):
  50. 1. Go to the [latest release](https://github.com/OWASP/secure-agent-playbook/releases/latest)
  51. 2. Download the zip for each skill you want (e.g. `web-security-review.zip`)
  52. 3. Upload it at **Organization settings > Skills > Upload a skill**
  53. > **Note:** The skills upload rejects zips with more than 200 files, so the full `secure-agent-playbook.zip` cannot be uploaded there — use the per-skill zips. To build them from a clone, run `scripts/package_skills.sh` and upload from `dist/skills/`.
  54. **Local development** — To test from a local clone instead of GitHub:
  55. ```
  56. /plugin marketplace add /path/to/agent-security-playbook
  57. /plugin install code-security-skills@agent-security-playbook
  58. /plugin install ai-security-skills@agent-security-playbook
  59. ```
  60. **Without Claude Code:**
  61. Reference plays directly as procedures for any AI agent or manual use:
  62. - Point your agent at a play: *"Follow the procedure in `plugins/ai-security-skills/plays/agent-security-audit.md`"*
  63. - Or use the plays as checklists for manual security reviews
  64. ## Skills Catalog
  65. ### Code & Infrastructure Security (`code-security-skills`)
  66. | Skill | What It Does | Say This | OWASP Ref |
  67. |-------|-------------|----------|-----------|
  68. | `code-review-security` | Security code review mapped to Top 10 + ASVS | "Review this code for security issues" | Top 10, ASVS |
  69. | `sca-audit` | Dependency CVE scanning with reachability analysis | "Scan my dependencies for CVEs" | A06:2021 |
  70. | `secrets-scan` | Detect hardcoded credentials and API keys | "Scan for hardcoded secrets" | CWE-798 |
  71. | `api-security-review` | API review against OWASP API Top 10 (2023) | "Review this API for security" | API Top 10 |
  72. | `web-security-review` | Web app review against OWASP Top 10 (2021) | "Review this web app for OWASP Top 10" | Top 10 |
  73. | `mobile-code-review` | Native Android/iOS source review against OWASP MASVS v2.1.0 | "Review this mobile app for security" | MASVS v2.1.0 |
  74. | `iac-security-review` | IaC security (Terraform, K8s, CloudFormation) | "Review this Terraform for security" | CIS Benchmarks |
  75. | `securability-engineering` | Generate inherently securable code (FIASSE v1.0.4) | "Generate secure code for..." | FIASSE v1.0.4 |
  76. | `securability-engineering-review` | Assess code securability (0-10 SSEM scoring) | "Assess the securability of this code" | FIASSE v1.0.4/SSEM |
  77. | `prd-securability-enhancement` | Harden PRDs/specs with ASVS coverage and FIASSE SSEM requirements before code is written | "Harden this PRD" / "Map features to ASVS" | ASVS, FIASSE v1.0.4 |
  78. | `security-guidance` | Auto-triggered ASVS guidance for security-sensitive code | *(auto-triggered)* | ASVS 5.0 |
  79. ### AI & Agent Security (`ai-security-skills`)
  80. | Skill | What It Does | Say This | OWASP Ref |
  81. |-------|-------------|----------|-----------|
  82. | `agent-security-audit` | Audit agent permissions, injection surfaces, data exfil paths | "Audit this agent's security" | LLM Top 10 |
  83. | `llm-risk-assess` | LLM app assessment against LLM Top 10 2025 | "Assess LLM risks for this app" | LLM Top 10 |
  84. | `agentic-ai-risk-assess` | Agentic app assessment against Top 10 Agentic 2026 | "Assess agentic AI risks" | Agentic Top 10 |
  85. | `mcp-server-review` | MCP server security review | "Review this MCP server" | LLM Top 10 |
  86. | `prompt-injection-test` | Prompt injection testing (Arcanum PI Taxonomy) | "Test for prompt injection" | LLM01 |
  87. | `multi-agentic-threat-model` | CSA MAESTRO 7-layer threat modeling | "Model threats for this multi-agent system" | CSA MAESTRO |
  88. | `ai-security-verification` | AI application verification against OWASP AISVS (13 categories) | "Verify this AI app against AISVS" | AISVS |
  89. ## Agents
  90. Agents are autonomous security specialists that invoke skills and produce structured reports. Each agent has a focused system prompt, scoped tool access, and preloaded skills. Use them individually or as a coordinated team.
  91. | Agent | Focus | Skills Invoked |
  92. |-------|-------|---------------|
  93. | `code-security-reviewer` | Code vulnerabilities, secrets, web security | code-review-security, secrets-scan, web-security-review |
  94. | `dependency-auditor` | Supply chain and dependency CVE risks | sca-audit |
  95. | `api-security-reviewer` | API security against OWASP API Top 10 | api-security-review |
  96. | `mobile-security-reviewer` | Native Android/iOS source against OWASP MASVS v2.1.0 | mobile-code-review |
  97. | `ai-security-assessor` | Agent configs, MCP servers, LLM app risks | agent-security-audit, mcp-server-review, llm-risk-assess, prompt-injection-test |
  98. | `security-team-lead` | Coordinates specialists, consolidates report | securability-engineering-review |
  99. **Standalone usage** — invoke any agent directly:
  100. ```
  101. "Use code-security-reviewer to review src/"
  102. "Use dependency-auditor to scan this project"
  103. ```
  104. **Team assessment** — with agent teams enabled, the team lead dispatches specialists in parallel and consolidates findings into a single report:
  105. ```
  106. "Run a full security assessment of this project"
  107. ```
  108. The team lead scopes the target, dispatches relevant specialists (skipping those whose focus area isn't present), deduplicates findings, identifies cross-domain risk chains, and produces a unified report using `templates/report.md`.
  109. > Agent teams requires `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1`. See [Claude Code docs](https://code.claude.com/docs/en/agent-teams) for setup. Individual agents work without this flag.
  110. ## Example Output
  111. Running `"Review src/auth/ for security issues"` on a Node.js/Express codebase produces findings like this:
  112. ```
  113. Security Code Review — src/auth/
  114. Scope: Node.js/Express, server-side, processes user credentials
  115. Findings: CRITICAL 1 | HIGH 2 | MEDIUM 1 | LOW 0
  116. ```
  117. ### [CRITICAL] SQL Injection in User Lookup
  118. - **CWE**: [CWE-89](https://cwe.mitre.org/data/definitions/89.html)
  119. - **OpenCRE**: [CRE-764-507](https://www.opencre.org/cre/764-507) — Injection Prevention
  120. - **OWASP Ref**: A03:2021 Injection
  121. - **Location**: `src/auth/login.js:42`
  122. - **Impact**: Attacker can extract the entire users table, bypass authentication, or execute arbitrary SQL via crafted `id` parameter
  123. - **Evidence**:
  124. ```js
  125. // src/auth/login.js:42
  126. const user = await db.query("SELECT * FROM users WHERE id=" + req.params.id);
  127. ```
  128. - **Remediation**: Use parameterized queries:
  129. ```js
  130. const user = await db.query("SELECT * FROM users WHERE id = $1", [req.params.id]);
  131. ```
  132. - **Confidence**: HIGH
  133. > Every skill produces structured findings with severity, CWE, evidence, and remediation code.
  134. > See [`examples/`](examples/) for complete sample assessment reports.
  135. ## Plays
  136. ### AI/Agent Security plays
  137. The differentiator — security procedures purpose-built for the AI agent era. Bundled inside the `ai-security-skills` plugin.
  138. | Play | What It Does |
  139. |------|-------------|
  140. | [agent-security-audit](plugins/ai-security-skills/plays/agent-security-audit.md) | Audit agent permissions, prompt injection surfaces, data exfiltration paths, guardrails |
  141. | [agentic-ai-risk-assess](plugins/ai-security-skills/plays/agentic-ai-risk-assess.md) | Assess agentic AI applications against OWASP Top 10 for Agentic Applications 2026 |
  142. | [ai-security-verification](plugins/ai-security-skills/plays/ai-security-verification.md) | Verify AI-driven applications against the OWASP AI Security Verification Standard (AISVS) |
  143. | [llm-risk-assess](plugins/ai-security-skills/plays/llm-risk-assess.md) | Assess LLM applications against OWASP Top 10 for LLM Applications |
  144. | [mcp-server-review](plugins/ai-security-skills/plays/mcp-server-review.md) | Review MCP server implementations for overpermissioning, injection, data exposure |
  145. | [multi-agentic-threat-model](plugins/ai-security-skills/plays/multi-agentic-threat-model.md) | Threat-model multi-agent systems using the CSA MAESTRO 7-layer framework and OWASP Multi-Agentic System Guide v1.0 |
  146. | [prompt-injection-testing](plugins/ai-security-skills/plays/prompt-injection-testing.md) | Test LLM apps against 18 attack techniques, 20 evasions, 13 intents |
  147. ### Code & Dependency Analysis plays
  148. Immediate, practical value for any codebase. Bundled inside the `code-security-skills` plugin.
  149. | Play | What It Does |
  150. |------|-------------|
  151. | [sca-audit](plugins/code-security-skills/plays/sca-audit.md) | Scan dependencies for known CVEs with reachability analysis |
  152. | [code-review-security](plugins/code-security-skills/plays/code-review-security.md) | Systematic security code review mapped to OWASP Top 10 and ASVS |
  153. | [secrets-scan](plugins/code-security-skills/plays/secrets-scan.md) | Detect hardcoded credentials, API keys, and tokens |
  154. | [api-security-review](plugins/code-security-skills/plays/api-security-review.md) | Review APIs against OWASP API Security Top 10 |
  155. | [owasp-top10-web-review](plugins/code-security-skills/plays/owasp-top10-web-review.md) | Web application review against OWASP Top 10 (2021) |
  156. | [mobile-code-review](plugins/code-security-skills/plays/mobile-code-review.md) | Native Android/iOS source code review against OWASP MASVS v2.1.0 |
  157. | [iac-security-review](plugins/code-security-skills/plays/iac-security-review.md) | Review Terraform, Kubernetes, CloudFormation against CIS benchmarks and cloud security best practices |
  158. | [securability-engineering-review](plugins/code-security-skills/plays/securability-engineering-review.md) | Assess code against FIASSE v1.0.4/SSEM securable attributes: Maintainability, Trustworthiness, Reliability, and Transparency |
  159. ### Planned
  160. - **Tier 2**: Threat modeling, ASVS verification, infrastructure hardening
  161. - **Tier 3**: WSTG testing checklist, DAST orchestration, attack surface mapping
  162. - **Tier 5**: SAMM maturity assessment, compliance mapping, aggregate reporting
  163. ## Architecture
  164. Three-layer design — every layer lives inside the plugin source folder so the marketplace install bundles everything the skills need:
  165. - **`plugins/<name>/agents/`** — Autonomous security specialists with focused system prompts, co-located inside each plugin. Each agent invokes one or more skills, operates in an isolated context, and produces structured reports. Can work solo or as a coordinated team.
  166. - **`plugins/<name>/skills/`** — Self-contained `SKILL.md` files following the [Agent Skills spec](https://agentskills.io/specification). Distributed through `plugins/code-security-skills/` and `plugins/ai-security-skills/`, registered in the marketplace via `.claude-plugin/marketplace.json`. Each skill summarizes a procedure and references its corresponding play inside the same plugin.
  167. - **`plugins/<name>/plays/`** — Full reference procedures with detailed checklists, tables, decision criteria, and examples. Skills reference these for comprehensive coverage.
  168. - **`plugins/<name>/templates/`** and **`plugins/<name>/data/`** — Output templates (`finding.md`, `report.md`) and OWASP reference datasets (FIASSE v1.0.4, ASVS v5.0, secure-code prompts) that skills load at runtime.
  169. Agents orchestrate, skills execute, plays provide the full procedure. Contributors edit plays. This means the playbook works with any AI agent (just point it at a play), while Claude Code users get plugin-based installation with agents and skills.
  170. ## OWASP Foundation
  171. All plays reference OWASP standards and datasets:
  172. - [OWASP Top 10](https://owasp.org/www-project-top-ten/) — Web application risks
  173. - [OWASP API Security Top 10](https://owasp.org/API-Security/) — API-specific risks
  174. - [OWASP Top 10 for LLM Applications](https://genai.owasp.org) — AI/LLM risks
  175. - [OWASP Agentic Security Initiative](https://genai.owasp.org/initiatives/agentic-security-initiative/) — Autonomous-agent risks (Top 10 for Agentic Applications)
  176. - [OWASP ASVS](https://owasp.org/www-project-application-security-verification-standard/) — Security verification requirements
  177. - [OWASP MASVS](https://mas.owasp.org/MASVS/) — Mobile Application Security Verification Standard
  178. - [OWASP AISVS](https://github.com/OWASP/AISVS) — AI Security Verification Standard
  179. - [OWASP WSTG](https://owasp.org/www-project-web-security-testing-guide/) — Testing methodology
  180. - [OWASP SAMM](https://owaspsamm.org) — Security program maturity model
  181. - [OWASP Cheat Sheet Series](https://cheatsheetseries.owasp.org) — Developer security guidance
  182. - [OWASP FIASSE](https://owasp.org/www-project-fiasse) — Framework for Inherently Adaptive and Securable Engineering (v1.0.4)
  183. ## Related Projects
  184. | Project | Relationship |
  185. |---------|-------------|
  186. | [OWASP Agent Skills Project](https://github.com/eoftedal/owasp-agent-skills-project) | Proactive ASVS 5.0 guidance for AI coding agents — helps agents **write** secure code. We use their ASVS reference data in `plugins/code-security-skills/data/asvs/`. Complementary: they guide code generation, we find vulnerabilities in existing code. |
  187. | [Securability Engineering](https://github.com/Securability-Engineering) | Securable code generation (OWASP FIASSE) and secure code requirements (ASVS) via spec file analysis and generation constraint (benchmarked and tuned) for various AI code generation tools. |
  188. | [Arcanum PI Taxonomy](https://github.com/Arcanum-Sec/arc_pi_taxonomy) | Prompt injection attack classification by Jason Haddix. Our `prompt-injection-testing` play is built on this taxonomy. CC BY 4.0. |
  189. | [OpenCRE](https://www.opencre.org) | Cross-standard requirement mappings (CWE, ASVS, WSTG, NIST 800-53). We use OpenCRE links in findings for multi-framework traceability. |
  190. ## Contributing
  191. See [CONTRIBUTING.md](CONTRIBUTING.md) for detailed guidelines.
  192. New plays should:
  193. - Solve one well-defined security task
  194. - Include clear trigger conditions (when should this play run?)
  195. - Follow a structured procedure with checkpoints
  196. - Produce findings using the in-plugin `templates/finding.md` format
  197. - Reference OWASP standards where applicable
  198. - Prefer existing tools (semgrep, trivy, osv-scanner, trufflehog) over reimplementing detection
  199. - Live inside the appropriate plugin's `plays/` folder so they ship with the marketplace install
  200. ## License
  201. Licensed under [Creative Commons Attribution 4.0 International](LICENSE.md) (CC-BY-4.0). See [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md) for attribution of upstream OWASP project content.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.