skills/mukul975/anthropic-cybersecurity-skills/testing-for-broken-access-control/Gen Agent Trust Hub
testing-for-broken-access-control
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external web server responses, which represents an indirect prompt injection attack surface.
- Ingestion points: The script
scripts/agent.pyprocesses HTTP response content (resp.text) and status codes in several testing functions. - Boundary markers: There are no boundary markers or instructions for the agent to ignore potentially malicious content within the server responses.
- Capability inventory: The skill utilizes network capabilities via the
requestslibrary and shell-based tools likecurlandffuf. - Sanitization: No input sanitization or validation is performed on the data received from external servers beyond status code checks and basic substring matching.
- [COMMAND_EXECUTION]: The skill utilizes shell commands and local scripts to perform network probing.
- Evidence: The
SKILL.mdfile contains instructions for executingffuf,curl, andbashloops to discover and test web endpoints.
Audit Metadata