The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill generates CI pipeline configurations that ingest and process potentially untrusted data from the local repository during automated builds.
Ingestion points: The generated GitHub Action workflow files (foundry-ci.yml) ingest data from requirements.txt and the tests/ directory during the build and test jobs.
Boundary markers: The provided templates do not include specific boundary markers or instructions to the environment to ignore embedded instructions within these processed files.
Capability inventory: The skill utilizes Bash capabilities to configure repository secrets via the GitHub CLI (gh) and suggests workflows that execute commands using pip, pytest, and ruff.
Sanitization: No explicit sanitization or integrity validation is performed on the repository contents before they are processed by the CI tools, which is standard for developer-centric CI setups but represents the primary surface for indirect injection if repository contents are compromised.

palantir-ci-integration