agent-ready
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from externally generated reports and server log files to drive its implementation workflow.\n
- Ingestion points: Pasted scorecards in user prompts and server-side log files as described in SKILL.md and references/verification.md.\n
- Boundary markers: No explicit delimiters or instructions to ignore embedded instructions are required for the ingested data.\n
- Capability inventory: Includes shell command execution (curl, npx, git, bash), file writing, and implementation logic across multiple scripts.\n
- Sanitization: No sanitization of the input data is described before it is used to determine code and configuration changes.\n- [EXTERNAL_DOWNLOADS]: Fetches benchmarking tools and scanner packages from remote sources to evaluate site readiness.\n
- Downloads and runs afdocs and is-agentic via npx as part of the discovery workflow.\n
- Clones the url-discovery-bench repository from the Mintlify GitHub organization for performance measurement.\n- [COMMAND_EXECUTION]: Utilizes shell utilities and scripts for system verification and log analysis.\n
- Executes a bundled bash script scripts/check-surfaces.sh which uses curl and awk to probe HTTP endpoints.\n
- Uses grep and awk commands to parse server access logs for AI agent activity patterns.\n- [DYNAMIC_EXECUTION]: Performs runtime installation and execution of external benchmarking code.\n
- Runs pip install on requirements from the cloned url-discovery-bench repository and executes its logic via python -m.
Audit Metadata