a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Review Infrastructure-as-Code (Terraform, CloudFormation, ARM/Bicep, Pulumi, Ansible, Helm) for security misconfigurations before deployment — public exposure, weak IAM, missing encryption, open networking, and hardcoded secrets. Use to shift-left and catch cloud/k8s misconfig
$ npx -y skills add jassics/awesome-claude-security --skill iac-security-review --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/iac-security-reviewContext preview
The summary Claude sees to decide when to auto-load this skill.
Review Infrastructure-as-Code (Terraform, CloudFormation, ARM/Bicep, Pulumi, Ansible, Helm) for security misconfigurations before deployment — public exposure, weak IAM, missing encryption, open networking, and hardcoded secrets. Use to shift-left and catch cloud/k8s misconfig
name: iac-security-review description: >- Review Infrastructure-as-Code (Terraform, CloudFormation, ARM/Bicep, Pulumi, Ansible, Helm) for security misconfigurations before deployment — public exposure, weak IAM, missing encryption, open networking, and hardcoded secrets. Use to shift-left and catch cloud/k8s misconfig at the code layer.
A triaged set of IaC security findings, each tied to the resource/line and a fix — so misconfigurations are caught before they reach a live environment.
1. **Public exposure** — resources open to the internet (security groups/NSGs/ firewall `0.0.0.0/0`, public buckets/storage, public IPs on sensitive hosts). 2. **Identity** — over-broad IAM policies, wildcard permissions, default/no roles, long-lived credentials defined in code, session duration raised beyond what the workload needs, and trust policies missing confused-deputy conditions (`aws:SourceArn`/`SourceAccount`/`ExternalId`) on third-party/cross-account roles. 3. **Encryption** — storage/volumes/databases without encryption at rest; TLS not enforced; default/unmanaged keys. 4. **Logging & monitoring** — audit logging/flow logs not enabled by the IaC. 5. **Secrets** — hardcoded passwords/keys/tokens in variables, defaults, or state (cross-ref `secrets-management-review`). 6. **Module/provider hygiene** — untrusted/unpinned modules, provider versions, drift between code and deployed state. 7. **Helm chart defaults** — a chart's `values.yaml` default (empty `resources`, `networkPolicy.enabled: false`, `ingress.tls: []`) silently becomes every `helm install`'s config, not just this deployment's — a distribution-scale version of the same "no safe default" risk, one layer up.
1. Identify the IaC type and scope; run an IaC scanner if available (Checkov, tfsec/Trivy, KICS, cfn-nag) or review the templates directly. 2. Triage findings: map each to the resource and confirm it's a real exposure (not a scanner false positive); note severity and blast radius. 3. Check for hardcoded secrets and sensitive values in variables/state. 4. Provide the corrected IaC snippet per finding.
A findings table: file:line · resource · misconfig · severity · fix (corrected snippet). Confirmed issues → `security-reporting:finding`. These map to the same controls `cloud-security` / `k8s-security` check at runtime.
Shift-left: fixing in IaC prevents the misconfig everywhere it's deployed and stops drift. Treat secrets in code/state as high severity. Pin and vet third-party modules — they run with your deploy credentials.
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…