a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Test an API's authorization — BOLA (object-level), BFLA (function-level), and BOPLA (property-level / mass assignment) — to confirm each request is authorized for the caller. Use to validate the top OWASP API risks on an authorized target.
$ npx -y skills add jassics/awesome-claude-security --skill api-authz-test --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/api-authz-testContext preview
The summary Claude sees to decide when to auto-load this skill.
Test an API's authorization — BOLA (object-level), BFLA (function-level), and BOPLA (property-level / mass assignment) — to confirm each request is authorized for the caller. Use to validate the top OWASP API risks on an authorized target.
name: api-authz-test description: >- Test an API's authorization — BOLA (object-level), BFLA (function-level), and BOPLA (property-level / mass assignment) — to confirm each request is authorized for the caller. Use to validate the top OWASP API risks on an authorized target.
Evidence on whether the API enforces authorization server-side for every object, function, and property against the authenticated caller — or whether a user can reach data/operations/fields they shouldn't.
an admin, with identifiable objects.
1. **BOLA (object level)** — as User A, swap object identifiers (path/query/body, IDs, UUIDs, GraphQL node IDs) to read or modify User B's objects. 2. **BFLA (function level)** — call admin/privileged operations as a normal user; try alternate methods/verbs, hidden endpoints, and GraphQL mutations not exposed to your role. 3. **BOPLA / mass assignment** — add privileged properties to a write (e.g. `"role":"admin"`, `"isVerified":true`) and check if they're accepted; inspect responses for over-exposed fields. 4. **Tenant isolation** — cross-tenant object access where multi-tenant.
1. Map objects, operations, roles, and the property model (use the schema/spec). 2. Run cases as the lower-privilege identity; capture request+response evidence. 3. Record per case: enforced / leaked / accepted, with evidence (redact secrets). 4. Identify the gap: ID trusted from request, no per-object/function/property authz, writable fields not allow-listed.
A results table: case · object/function/property · result · evidence · remediation (deny-by-default authz per object/function/property tied to identity; field allow-listing). Confirmed issues → `security-reporting:finding` (high+ for cross-user/tenant access).
BOLA + BFLA are the #1 and #5 API risks and cause most API breaches. Authorization must be enforced server-side per request — never infer it from the client-supplied ID, role, or tenant. Same class as web `access-control-test`.
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…