The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates external data into the agent's context without sanitization or boundary markers.
Ingestion points: Knowledge is ingested from files located in the .agents/learnings/, .agents/patterns/, .agents/research/, and .agents/retros/ directories (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the injected content as data rather than instructions.
Capability inventory: The skill utilizes the ao CLI tool and standard file-reading capabilities to process these artifacts.
Sanitization: No evidence of validation, escaping, or filtering of the file content exists prior to injection.
[COMMAND_EXECUTION]: The skill executes shell commands to discover and process knowledge artifacts.
Evidence: Uses ls -lt and head to find recent files, and the ao CLI tool to perform context-aware injection.

inject