a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Assess and establish build artifact provenance and integrity — SLSA provenance level, signing/attestation (Sigstore/cosign, in-toto), and verification at deploy/admission. Use when you need to prove an artifact (container image, package, binary) came from the expected source and
$ npx -y skills add jassics/awesome-claude-security --skill artifact-provenance-verification --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/artifact-provenance-verificationContext preview
The summary Claude sees to decide when to auto-load this skill.
Assess and establish build artifact provenance and integrity — SLSA provenance level, signing/attestation (Sigstore/cosign, in-toto), and verification at deploy/admission. Use when you need to prove an artifact (container image, package, binary) came from the expected source and
name: artifact-provenance-verification description: >- Assess and establish build artifact provenance and integrity — SLSA provenance level, signing/attestation (Sigstore/cosign, in-toto), and verification at deploy/admission. Use when you need to prove an artifact (container image, package, binary) came from the expected source and build, untampered.
A clear picture of an artifact's provenance maturity and a path to "only verified, attested artifacts are deployed" — so a tampered or unknown-origin artifact can't reach production.
Map the build/artifact flow to a **SLSA** build level and identify the next gap:
non-forgeable. This is the meaningful bar for tamper resistance.
1. **Inventory trust signals** — is the artifact signed? By what identity (keyless/OIDC via Sigstore, or a managed key)? Is there a provenance attestation (in-toto / SLSA) linking artifact ↔ source commit ↔ builder? 2. **Verify the chain** — source → build → artifact → registry → deploy. Confirm the signing identity and provenance are checked, not just present. 3. **Find the enforcement gap** — is verification *required* at admission/deploy (e.g., cosign verify in an admission controller / policy gate), or merely available? Unverified-but-signed is theater. 4. **Recommend** — signing in CI (keyless preferred), provenance generation, verification policy at the gate, and key/identity trust roots + revocation.
A provenance assessment: current SLSA level · signing/attestation state · where verification is (and isn't) enforced · prioritized steps to the next level. Pair with `pipeline-integrity-review` for the build side and `security-diagramming` for the chain-of-custody diagram.
Signing without enforced verification buys nothing — the control is the *gate that rejects unverified artifacts*. Keyless (OIDC/Fulcio) signing avoids long-lived key management and is the modern default. Provenance must bind artifact → exact source commit → builder, or it can't prove "this came from that build."
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…