The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest data from external feedback files and use it to rewrite its own instructions and other skill files, creating an attack surface where malicious input could influence agent behavior.
Ingestion points: Processes files matching .claude/feedback/retro-*.md in SKILL.md.
Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions within the feedback data.
Capability inventory: The skill utilizes file system operations (ls, mv, mkdir) and markdown editing capabilities to modify existing skills.
Sanitization: No sanitization or validation logic is defined to prevent instructions embedded in feedback from being promoted into permanent skill instructions.
[COMMAND_EXECUTION]: The skill uses shell commands for file management and includes a bash script template for generating weekly digests.
Evidence: Uses ls -la .claude/feedback/retro-*.md, mkdir -p .claude/feedback/archive, and mv .claude/feedback/retro-*.md .claude/feedback/archive/ to manage files.
Evidence: Provides a script template weekly-feedback-digest.sh that uses find and grep to process feedback files.

skill-improver