The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes various project documents and git logs to decide which work has landed and which files to delete. This creates a surface where malicious instructions embedded in project docs could influence the agent.
Ingestion points: Reads PLAN.final.md, EXECUTION.md, ACCEPTANCE.md, HANDOFF.md, and documents in docs/<feature>/.
Boundary markers: The skill does not specify the use of delimiters or ignore-instructions when reading these files.
Capability inventory: Includes file reading, writing, and deletion, along with executing git commands, grep, and project "doc generators".
Sanitization: No explicit sanitization or validation of external file content is described.
[COMMAND_EXECUTION]: The skill executes shell commands like git log, git fetch, and grep to analyze the repository state. It also calls for running "doc generators", which are project-specific executable tools.
[FILE_SYSTEM_MODIFICATION]: The skill's primary function involves deleting multiple planning files once a feature is complete. Although it requires user approval before performing deletions, the selection of files for removal is determined by parsing the project's own documentation.

close-plan