The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands (mkdir, mv) to organize files and uses Python's subprocess module in scripts/validator.py to interact with the local git binary. These operations are used for repository status checks, commit history inspection, and committing moves.
[INDIRECT_PROMPT_INJECTION]: The scripts/validator.py script parses content from user-generated Markdown artifacts (intents, specs, plans, and reports). The agent is instructed to follow the output and failure reasons generated by this script, creating a surface where malicious instructions in the artifacts could theoretically influence the agent's workflow.
Ingestion points: scripts/validator.py reads intent.md, spec.md, plan.md, and report.md from the intent/ directory.
Boundary markers: The script performs structural validation and digest verification, but does not provide explicit prompt delimiters when reporting errors to the agent.
Capability inventory: The skill can move directories and perform git commits.
Sanitization: The script uses regex patterns to validate the structure and content of the artifacts but does not sanitize text before it is returned as part of a validation failure message.
[SAFE]: The skill's operations are confined to the local file system and the project's Git repository. Path traversal is mitigated by validating the slug against a strict alphanumeric regex pattern (^[A-Za-z0-9][A-Za-z0-9._-]*$). No network access or external downloads were detected.

sdlc-archive