The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts using the python interpreter via the command line. Specifically, it runs scripts/self_improvement/lesson_capture.py to mine logs and diagnose performance issues, and scripts/quality_gates/golden_check.py to verify gate changes.
[DYNAMIC_EXECUTION]: The skill instructs the agent to modify source code in scripts/quality_gates/seo_lint.py by adding new regex rules or Tier entries. It also modifies evals/golden/manifest.json to add new test cases that are subsequently executed.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which serves as a vulnerability surface for indirect prompt injection.
Ingestion points: The agent reads data from data/learning/gate_runs.jsonl, content_log.json, and the text of underperforming pieces to generate diagnoses and lessons.
Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted log data or content from the agent's core instructions.
Capability inventory: The skill has the ability to write to files (memory/*.md, scripts/*.py) and execute shell commands (python scripts/...).
Sanitization: The instructions do not specify any validation or sanitization steps for the data extracted from logs or pieces before it is used to generate diagnoses or new gate rules.

kai-retro