The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The skill executes arbitrary shell commands defined in local project configuration files to validate the build and test status.
Evidence: The bash snippets in 'Check 2' and 'Check 3' execute npm test, npm run build, pytest, cargo test, and go test based on the existence of project manifest files like package.json or Cargo.toml.
Context: While this is the primary purpose of a 'validate-delivery' skill, it means the agent will execute any code defined in the target repository's test or build scripts.
[PROMPT_INJECTION] (LOW): The skill contains an indirect prompt injection surface by parsing external task descriptions into instructions without sanitization.
Ingestion points: The extractRequirements function parses task.description for bulleted and numbered lists (SKILL.md, Check 4).
Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands within the extracted requirements.
Capability inventory: The skill has the capability to execute shell commands (npm, pytest, etc.) and modify workflow state via workflowState.completePhase.
Sanitization: Absent. Requirements are extracted via regex and passed directly to a verification function.
[DATA_EXFILTRATION] (SAFE): No hardcoded credentials, sensitive file access (beyond project manifests), or unauthorized network requests were detected.

validate-delivery