a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Run or ingest static application security testing (SAST) results on a codebase, triage them to remove false positives, and confirm the real issues with code evidence and remediation. Use when reviewing source code for security flaws or cleaning up noisy scanner output.
$ npx -y skills add jassics/awesome-claude-security --skill sast-review --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/sast-reviewContext preview
The summary Claude sees to decide when to auto-load this skill.
Run or ingest static application security testing (SAST) results on a codebase, triage them to remove false positives, and confirm the real issues with code evidence and remediation. Use when reviewing source code for security flaws or cleaning up noisy scanner output.
name: sast-review description: >- Run or ingest static application security testing (SAST) results on a codebase, triage them to remove false positives, and confirm the real issues with code evidence and remediation. Use when reviewing source code for security flaws or cleaning up noisy scanner output.
A triaged SAST finding set: confirmed, code-evidenced issues mapped to CWE and ranked — with the false positives filtered out.
1. **Scope & scan** — identify languages/frameworks; run an appropriate SAST tool (e.g. Semgrep/CodeQL or the project's configured scanner), or ingest existing results. Note coverage and any excluded paths. 2. **Triage each finding** — open the flagged code and trace data flow from source (untrusted input) to sink. Classify: true positive / false positive / needs runtime confirmation. Watch for sanitizers/validation the tool missed. 3. **Confirm impact** — for true positives, identify the vulnerability class (CWE), reachability, and exploitability; mark which need dynamic confirmation (`web-app-security` / `api-security`). 4. **Score & rank** — severity (`security-reporting:cvss`) weighted by reachability; prioritize reachable, high-impact issues. 5. **Remediate** — give the specific code-level fix (parameterize, encode, validate, safe API) — and a secure pattern to prevent recurrence.
A triaged findings table: id · CWE · file:line · class · reachability · TP/FP · severity · remediation. Confirmed issues → `security-reporting:finding`.
SAST is high-recall, low-precision — the value you add is triage. Always trace source→sink before accepting a finding, and flag what needs dynamic confirmation rather than over-claiming. Tune rules to cut recurring false positives. For Python or React/JS specifically, `secure-coding:safe-function-lint` covers a smaller, curated banned-function rule pack (CWE/ASVS-cited) *and* can enforce it via installed git hooks — this skill stays the broader, multi-language, scan-any- codebase pass; the two aren't mutually exclusive.
A Claude Code plugin marketplace for the full cybersecurity & GenAI-security lifecycle — from recon and threat modeling to detection engineering, GRC, and CISO-level strategy. A pentester knows which OWASP test bends a broken-access-control endpoint.
Repo: jassics/awesome-claude-security
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Test the agent execution harness/runtime itself — LangChain/LangGraph, AutoGen, CrewAI, custom ReAct-style loops, or computer-use/browser-use agents — for…
Assess an autonomous / tool-using AI agent for security end-to-end: tool privileges, autonomy and approval boundaries, excessive agency, memory/state…
Test what an AI agent will actually do without human confirmation, including under injected-goal / prompt-injection scenarios, to validate its autonomy and…
Review the security of MCP (Model Context Protocol) servers/clients an agent uses: server trust tier, tool/resource description and result poisoning,…
Inventory the tools/functions an AI agent can call and audit their privileges, side effects, and approval requirements to find excessive-agency and…