The Agent Skills Directory

[SAFE]: The skill is designed to perform in-session re-orientation by summarizing conversation history into a structured 7-block schema. It does not request or use any external tools.
[SAFE]: The instructions in SKILL.md and references/seven-block-schema.md explicitly prohibit writing data to disk, ensuring all output remains within the ephemeral chat session.
[SAFE]: Technical analysis of the provided Python test scripts (scripts/test_readmes.py, scripts/test_seven_block_schema.py, scripts/test_skill_md.py) confirms they are benign unit tests. They use standard libraries like pathlib, re, and PyYAML to verify the presence of required headings, triggers, and principles within the documentation.
[SAFE]: No evidence of prompt injection, obfuscation, or data exfiltration was detected. The skill follows established best practices for agent-to-user communication, such as the 'Synthesis-check' and 'plain-language' principles.

recap-state