The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several local bash scripts to automate the retrospective workflow, including lesson retrieval and formatting.
Evidence: scripts/retro-lessons.sh performs text manipulation and file management operations using standard utilities like grep, sed, and cat.
[PROMPT_INJECTION]: The skill provides snippets to be added to instruction-bearing files like AGENTS.md and .cursorrules to modify agent behavior and ensure the retrospective tool runs automatically.
Evidence: references/auto-trigger.md contains an 'Auto-Retro Rule' intended to persist agent instructions across sessions.
Ingestion points: ~/.agents/lessons/LESSONS.md (read by scripts/retro-lessons.sh in the inject subcommand).
Boundary markers: Lessons are enclosed in ## Relevant Lessons blocks and tagged with  HTML comments.
Capability inventory: The skill is capable of executing shell scripts and modifying local configuration files based on the content of its logs.
Sanitization: Lesson content is extracted and placed into the context without specific escaping, which could allow instructions in the logs to be interpreted by the LLM.

retro