The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages the creation and execution of shell scripts (such as .sh or .ps1) within the test_scripts/ directory to facilitate environment setup and testing procedures.
[REMOTE_CODE_EXECUTION]: The skill generates and executes local scripts incorporating user-defined logic and also invokes the codeplain CLI tool, which communicates with a remote API at https://api.codeplain.ai for code rendering.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. User-provided information collected during the multi-phase QA interview is used to author specification files and testing scripts without explicit sanitization.
Ingestion points: User input captured during Phases 1, 2, and 3 via the AskUserQuestion tool and free-form prompts.
Boundary markers: No specific delimiters or "ignore" instructions are employed to isolate user-provided content from the generated specification structures or script templates.
Capability inventory: The agent has permissions to write files to the workspace and execute generated shell scripts and the codeplain CLI.
Sanitization: The workflow relies on iterative manual review by the user and internal automated review tools rather than programmatic sanitization or escaping of external content before its use in executable contexts.
[EXTERNAL_DOWNLOADS]: During the rendering and validation process, the skill interacts with external services hosted on the api.codeplain.ai domain, which is a vendor-owned resource.

forge-plain