The Agent Skills Directory

[PROMPT_INJECTION]: The skill describes a process for reading and evaluating external content that could contain indirect prompt injections designed to manipulate the agent's behavior.
Ingestion points: The agent is instructed to ingest data from GitHub release notes via 'gh api', external 'CHANGELOG.md' files, and Helm chart value diffs (Phase 2 and 3).
Boundary markers: There are no instructions for the agent to wrap this external content in delimiters or to ignore embedded instructions that might appear in the changelog text.
Capability inventory: The skill allows for significant system interaction, including building Docker images, running test suites (pytest, go test, npm test), and performing dry-run deployments (kubectl apply, terraform plan).
Sanitization: The skill does not implement any validation or sanitization of the external text strings before the agent processes them.
[COMMAND_EXECUTION]: The skill guides the agent through the execution of numerous standard development and operations CLI tools. These include package managers (uv, npm, cargo, bundle, go), infrastructure tools (terraform, pulumi, helm, kubectl), and search utilities (ripgrep). These operations are essential to the skill's purpose and are documented clearly.
[EXTERNAL_DOWNLOADS]: The skill relies on official package registries and repositories to discover and download dependency updates. While these involve fetching external code, they are standard operations performed through trusted package management systems (e.g., pip-audit, npm-check-updates, cargo-outdated).

dependency-updates