The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides shell snippets for the agent to execute during the implementation phase. These commands generate a project-specific identifier using cksum and awk to create or remove a temporary flag file in the system's temporary directory (/tmp or $TMPDIR). This is used to coordinate behavior with other installed plugins.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process the content of project-specific instruction files (e.g., CLAUDE.md, AGENTS.md, and MEMORY.md) to perform cleanup and restructuring.
Ingestion points: The agent reads the contents of the directory tree, instruction files, and @-import referenced files as part of the Phase 1 Audit.
Boundary markers: The instructions do not implement explicit delimiters or boundary markers to isolate the content of analyzed files from the agent's core instructions.
Capability inventory: The skill uses grep for verification and performs file system write operations to restructure instructions and manage the temporary coordination flag.
Sanitization: There is no explicit sanitization or filtering logic applied to the content moved between files, as the agent is expected to follow a user-approved restructuring plan.

instruction-cleanup