The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides utility scripts scripts/init-session.sh and scripts/check-complete.sh. These scripts are used to initialize local markdown templates and verify task completion through simple text parsing with grep. They use standard, benign shell commands and do not interact with sensitive system files or network resources.
[INDIRECT_PROMPT_INJECTION]: The skill encourages the agent to maintain context by reading and writing to files like task_plan.md and findings.md which may store data from external research.
Ingestion points: Files task_plan.md, findings.md, and progress.md are read by the agent to recover state during long tasks or after context resets.
Boundary markers: The skill provides predefined markdown templates with clear headers and sections to organize data.
Capability inventory: The skill utilizes basic file system operations and the provided shell scripts for context management.
Sanitization: No specific sanitization logic is implemented for external data, but the workflow encourages manual verification and structured logging to mitigate potential injection risks.
[REMOTE_CODE_EXECUTION]: Installation instructions refer to standard package management via npx and a public GitHub repository for the skill itself. No suspicious remote code execution patterns or downloads from untrusted sources were identified.

planning-with-files