The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies and processes numerous external project files, creating an indirect prompt injection surface where untrusted data could influence agent behavior.
Ingestion points: Processes discovery.md, benefit-metric.md, artefacts/[feature]/nfr-profile.md, .github/architecture-guardrails.md, and various story/epic definition files.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the parsed content.
Capability inventory: The skill is capable of executing the local scripts/validate-trace.sh script and performing write operations to .github/pipeline-state.json.
Sanitization: There is no evidence of sanitization, escaping, or validation of the external file content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on and provides instructions for the execution of a local shell script (scripts/validate-trace.sh) within CI environments (GitHub Actions, Jenkins, GitLab CI, and Azure Pipelines).
Evidence: The skill documentation includes snippets such as bash scripts/validate-trace.sh --ci and sh 'bash scripts/validate-trace.sh --ci' for various CI platforms.
Context: While the script is stored locally within the repository, its execution is a core dependency of the skill's functionality.

trace