The Agent Skills Directory

[PERSISTENCE]: The skill's primary function is to register hooks in .cursor/hooks.json. These entries ensure that specific shell scripts are executed automatically by the editor environment across different sessions whenever lifecycle events (session start, stop, or tool use) occur.
[DYNAMIC_EXECUTION]: The skill generates several shell scripts, including daily-end-reminder.sh and git-activity-buffer.sh, and writes them to the .groove/hooks/cursor/ directory. It uses chmod +x to make these generated files executable by the system.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by ingesting data that may be presented to the agent in future sessions.
Ingestion points: The git-activity-buffer.sh and block-managed-paths.sh scripts read tool invocation metadata from standard input. The session-capture.sh script collects local git logs and diffs.
Boundary markers: The generated pending-capture.md file, which is intended for agent review in subsequent sessions, does not use specific delimiters or instructions to ignore potentially malicious content embedded in git commit messages or diffs.
Capability inventory: The skill has the ability to write files and configure automated shell command execution via editor hooks.
Sanitization: While scripts use grep and python3 for basic parsing, the raw output of git commands is written directly to markdown files processed by the agent.
[COMMAND_EXECUTION]: The skill configures the environment to execute local scripts from other components, such as groove-utilities-prime.sh and check.sh, located within the .agents/skills/ directory.
[TIME_DELAYED_CONDITIONAL]: The version-check.sh hook includes logic to only execute its primary payload once every 3600 seconds (one hour), gating its execution based on a timestamp stored in a local cache file.

groove-admin-cursor-hooks