The Agent Skills Directory

[SAFE]: The skill interacts with the user's task board via the local tsk CLI tool. It follows security best practices by requiring the agent to wait for explicit user approval before editing or adding tasks and by refusing to run installers or package managers without a direct request.\n- [COMMAND_EXECUTION]: The skill utilizes the tsk CLI to manage tasks. This usage is confined to specific, non-administrative subcommands and includes error handling based on CLI exit codes to ensure reliable execution.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external task data and local source code, creating an indirect prompt injection surface. 1. Ingestion points: Task notes retrieved via tsk list and local project files, documentation, and tests (SKILL.md). 2. Boundary markers: No specific delimiters are defined for untrusted content. 3. Capability inventory: Task creation and modification via tsk add and tsk edit. 4. Sanitization: The mandatory 'propose, then write' rule acts as a human-in-the-loop control, preventing the autonomous execution of potentially malicious instructions embedded in the processed data.

tsk-cli