The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell scripts (ensure-ready.sh) and PowerShell scripts (ensure-ready.ps1) to verify the execution environment and project readiness.
[COMMAND_EXECUTION]: Frequently invokes the hamster CLI tool with subcommands such as status, sync, and ship to interact with the project's state and data.
[INDIRECT_PROMPT_INJECTION]: The skill reads and analyzes brief.md and task files in the .hamster/ directory, which could contain instructions intended to influence the agent's behavior during analysis modes. \n
Ingestion points: Reads content from .hamster/${account}/briefs/${slug}/brief.md and multiple associated task markdown files. \n
Capability inventory: The skill can execute hamster CLI commands that sync data or transition project status based on analysis. \n
Boundary markers: No specific delimiters or safety instructions are used to separate user-controlled task content from the agent's core instructions. \n
Sanitization: Content from task bodies is extracted and processed without specific sanitization or filtering.
[DYNAMIC_EXECUTION]: Employs Python heredocs within shell blocks to dynamically execute logic for parsing and validating the .hamster/.state.json file during account resolution.
[PRIVILEGE_ESCALATION]: The skill's documentation instructs the agent or user to re-run readiness checks with elevated permissions (e.g., sudo) if environment restrictions prevent access to necessary system credentials.

plan-hamster