The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the codebase and documentation files which could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: The agent reads documentation files (e.g., docs/devdocs/*.md) and scans source code files to detect implementation progress and deviations.
Boundary markers: There are no specific delimiters or "ignore instructions" wrappers defined for the content read from the workspace files before it is processed by the agent.
Capability inventory: The agent is granted Bash permissions to execute project-specific test commands and Write permissions to update documentation files.
Sanitization: The skill incorporates a safety constraint requiring explicit user confirmation via the AskUserQuestion tool before performing high-risk document updates or filesystem writes, which serves as a primary mitigation against unauthorized actions.

sync