The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructs the agent to parse a JSON response from the pocketto-pi CLI and execute every command listed in the next_actions array. This design allows for the execution of arbitrary shell commands if the CLI output is altered or malicious.\n- [COMMAND_EXECUTION]: The agent is directed to run various shell commands, including git show for commit metadata and npx for executing package-based tools.\n- [REMOTE_CODE_EXECUTION]: By utilizing npx -y pocketto-pi, the skill fetches and executes external code from the NPM registry at runtime, introducing potential supply chain vulnerabilities.\n- [EXTERNAL_DOWNLOADS]: The skill relies on npx to download the pocketto-pi package, which is an external dependency not pinned to a specific version or hash in the instructions.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from log.json, reviews/*.json, and phase-notes-*.json to build reports and make state transition decisions.\n
Ingestion points: log.json, reviews/*.json, and phase-notes-*.json files within the repository.\n
Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between its core logic and the content of these processed files.\n
Capability inventory: Access to shell command execution (npx, git) and the ability to write to the local filesystem (closeout.md).\n
Sanitization: The skill lacks validation or sanitization routines for the content of the JSON artifacts it processes.

pocket-closing