The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands to resolve paths, manage files, and verify code implementations through external build tools.
Evidence: Uses git rev-parse --show-toplevel, ls -1, make lint, and make test to interact with the project repository.
Context: These commands are used to identify the project root and execute the repository's native testing suites.
[PROMPT_INJECTION]: The skill is designed to ingest and execute instructions from external markdown files, creating a surface for indirect prompt injection.
Ingestion points: Reads implementation instructions from plan.md files located in the Obsidian vault at engineering/<project>/workplans/<plan>/plan.md.
Boundary markers: Absent. The agent is instructed to read the plan and "implement it properly" without specific delimiters or instructions to ignore embedded malicious prompts.
Capability inventory: The skill has extensive capabilities, including writing to the local repository, writing to the vault, and executing arbitrary shell commands defined in the project's Makefile.
Sanitization: Absent. There is no mention of validating or filtering the contents of the plan files before processing them.
[DATA_EXFILTRATION]: The skill performs automated network operations by pushing local content to a remote Git repository.
Evidence: Executes git -C "$V" push to synchronize the Obsidian vault (defaulting to $HOME/Documents/obsidian/obsidian) after updates.
Context: While intended for synchronization, this automates the transmission of local vault data to an external remote server.

ai-execute-plan