The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from various repository files to determine its deployment logic, which represents a potential attack surface if those files are maliciously modified.
Ingestion points: Reads configuration and instructions from docs/agents/code-host.md, docs/agents/workflow.md, docs/agents/issue-tracker.md, and repository-specific PR templates.
Boundary markers: The skill specifies that workflow.md is authoritative, providing a logical boundary, but does not define technical delimiters for the ingested content.
Capability inventory: Performs Git operations (commit, push, merge, branch management), interacts with code host APIs via CLI (e.g., gh), and manages ticket states.
Sanitization: No specific sanitization or validation logic is described for the text content read from the documentation files.
[REMOTE_CODE_EXECUTION]: The skill references and loads external dependencies (code-cleanup, code-review, pr). However, these are identified as part of the official 'Snappedly Skills' suite and are managed through the platform's standard skill invocation mechanism, which is a safe practice for modular skill design.

deploy