The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The github_workflow_fix workflow identifies and processes external data which could contain adversarial content.
Ingestion points: The skill retrieves GitHub Action job logs using gh run view <run-id> --log-failed in references/source-commands.md.
Boundary markers: Instructions do not specify boundary markers or delimiters for the ingested log content.
Capability inventory: The skill is capable of executing git push, gh pr create, and modifying files within a git worktree as defined in references/source-commands.md.
Sanitization: There is no evidence of sanitization or filtering applied to the logs before they are analyzed for root-cause diagnosis and automated fixes.
[COMMAND_EXECUTION]: The skill utilizes several CLI tools to perform its core functions, including git, GitHub CLI, and Node.js utilities.
Evidence: Executes git for branch management, gh for releases and workflow management, and npx bumpp for versioning in references/source-commands.md.
Safety context: The skill's execution rules in SKILL.md require explicit user confirmation before any destructive or external side effects (like push or release) are performed.
[EXTERNAL_DOWNLOADS]: The integrity check workflow performs network operations to verify resource availability.
Evidence: Uses curl -sI to validate icon URLs hosted on img.jsdelivr.com (referencing LobeHub repository content) in references/source-commands.md.
Context: These network operations are limited to HTTP HEAD requests for status code verification.

git-workflows