The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses the eval command to execute shell strings for repository installation ($install) and verification ($verify). These strings are derived from the repository's configuration file (.tituskirch-skills.json) and the detected lockfile type.
Evidence: Found in REFERENCE.md within the "Verify the PR's head" section for both GitHub and GitLab recipes, where fetched strings are executed via shell evaluation.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from external pull/merge requests to drive its logic.
Ingestion points: Pull request titles and descriptions retrieved via gh pr view and glab mr list (documented in SKILL.md Workflow Steps 2 and 3).
Boundary markers: The skill does not use specific delimiters or instructions to ignore embedded commands when processing PR content, relying solely on author verification as a security boundary.
Capability inventory: The skill possesses capabilities including merging PRs (gh pr merge/glab mr merge) and executing shell commands via the verify step.
Sanitization: No sanitization or escaping of ingested PR content is performed before it is used to determine update tiers or update types.

merge-deps