The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from specification files which serves as a surface for indirect prompt injection.
Ingestion points: Reads requirements and task plans from Markdown files located in the Docs/In-Progress/ directory.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to isolate content from the processed specification files.
Capability inventory: The workflow can execute shell commands (git, gh, python3), modify the local filesystem, and initiate network-based deployment/PR tasks via delegated skills.
Sanitization: There is no explicit sanitization or validation logic for the text ingested from the specification documents before it is used to guide the agent's work.
[COMMAND_EXECUTION]: The skill uses shell-based commands for validation and workflow automation.
Evidence: Employs python3 -c for YAML syntax checking and sed for line-based file inspection.
Validation: The Python command specifically uses yaml.safe_load, which is the correct secure practice to prevent arbitrary code execution during YAML parsing.
Workflow: Uses git and the GitHub CLI (gh) for branching, committing, and pull request management as part of the standard development cycle.

implement-and-review-loop