agent-decision-receipts
Warn
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
openagentontologypackage viapip. This library is not from a trusted organization or well-known service. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted agent metadata to build manifests, creating a surface for injection. 1. Ingestion points: Metadata is passed via CLI arguments to
scripts/build_action_manifest.py. 2. Boundary markers: No explicit delimiters are used to wrap the ingested content. 3. Capability inventory: Thescripts/build_action_manifest.pyscript possesses file-write capabilities. 4. Sanitization: The script includes a check for ASCII safety but does not perform content validation. - [COMMAND_EXECUTION]: The skill instructs the agent to run shell commands for installing dependencies, writing files, and executing Python one-liners.
Audit Metadata