The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a vendor-provided local bash script (source-registry.sh) to manage the project's source registry and calculate entity staleness.
[EXTERNAL_DOWNLOADS]: In its 'Check' mode, the agent fetches content from user-specified external URLs to compare content hashes and identify changes or drift in source data.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes untrusted data from external sources while having access to system capabilities.
Ingestion points: The agent reads and processes data fetched from external URLs during the source drift check (SKILL.md).
Boundary markers: The instructions lack explicit boundary markers or directives to ignore instructions that might be embedded within the retrieved external content.
Capability inventory: The agent is authorized to use the Bash tool for script execution and has Write and Edit permissions for project JSON files.
Sanitization: There is no mention of sanitization, filtering, or validation performed on the external content before the agent processes it.

portfolio-lineage