The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external sources.
Ingestion points: The skill instructions in SKILL.md require the agent to read external changelogs, release notes, and GitHub issues.
Boundary markers: There are no instructions to use boundary markers or to ignore instructions embedded within the external content.
Capability inventory: The skill has the capability to execute shell commands for building, testing, and Git operations (push/commit) across various environments (Node.js, Rust, Go, Python, Java, Docker, Flux).
Sanitization: No sanitization or escaping mechanisms are defined for the data retrieved from external sources.
[COMMAND_EXECUTION]: The skill performs extensive command-line operations to detect stacks, check for updates, and validate changes.
Evidence: The skill uses numerous CLI tools including npm, pnpm, yarn, bun, cargo, go, uv, poetry, pip, mvn, gradle, docker, flux, skopeo, and gh as described in SKILL.md and scripts/discover.sh.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to retrieve package metadata and repository indexes.
Evidence: The scripts/discover.sh script uses curl to fetch Helm index.yaml files and uses registry-specific tools like skopeo and npm to query for versions. These operations target well-known package registries and URLs defined in the project's configuration.

upgrade-deps