The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill's primary function is to ingest and follow instructions from manifest files, execution logs, and reference files to perform tasks.
Ingestion points: Manifest files (passed via $ARGUMENTS), execution log files, and behavioral mode files (e.g., references/execution-modes/thorough.md).
Boundary markers: The instructions do not define boundary markers or 'ignore' commands to isolate external manifest data from the agent's core instructions.
Capability inventory: The skill uses Bash (shell execution), Read, Write, and Edit tools, and calls other skills such as /verify, /define, and /escalate.
Sanitization: There is no evidence of sanitization or validation of the content within the manifest files before execution.
[COMMAND_EXECUTION]: The skill is designed to perform work through shell commands (Bash) and file system modifications, including navigating between different repositories using absolute paths.

do