The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git log -1 --format=%cI <path> to retrieve the timestamp of the last modification for existing skill files. This is a standard operation used to limit the scope of history mining to recent changes.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from past conversation logs to generate new system instructions.
Ingestion points: It reads transcript files located in the agent-transcripts/ directory of the active workspace (SKILL.md).
Boundary markers: The skill does not define specific technical delimiters to distinguish between conversation data and potential instructions within the transcripts during the mining phase.
Capability inventory: The skill has the ability to write and update files via the create-skill tool and perform git operations (worktrees, commits, PRs).
Sanitization: It relies on a human-in-the-loop review process ('vibe-check') and a prose refinement step using the 'unslop' skill to ensure the generated content is appropriate and accurate.

automate-me