The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to manage version control and environment identification.
Evidence includes executing git status, git commit, and git push for synchronization.
The skill uses $env:COMPUTERNAME (PowerShell) or hostname (Unix-like) to retrieve the machine identifier for logging purposes.
[INDIRECT_PROMPT_INJECTION]: The skill summarizes dialogue history to update handoff and roadmap files, creating a vulnerability surface where instructions embedded in previous conversation turns could influence the generated summary.
Ingestion points: Dialogue history and context are read from the session to generate progress reports (e.g., in handoff.md and Obsidian logs).
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands during the summarization process.
Capability inventory: The skill has write access to the filesystem (e.g., AGENTS.md, handoff.md, Obsidian vault files) and network capability via git push to synchronize changes to remote repositories.
Sanitization: No specific sanitization or escaping mechanisms are mentioned for processing the dialogue content before it is written to files or Git commits.

shutdown