The Agent Skills Directory

[SAFE]: The skill performs legitimate workflow management tasks, including reading work item statuses and posting comments to project management platforms like Jira and Linear.- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting untrusted data from external sources.
Ingestion points: Data is retrieved via getWorkItem from tracker APIs and through local shell commands like git log and git show as described in SKILL.md.
Boundary markers: The skill refers to rendering via external templates (prompts/coding/templates/impl-log-template.md), but does not explicitly define sanitization or boundary markers within this skill file.
Capability inventory: The skill possesses the ability to perform network writes (via comment and publishRecord) and file writes to the local documentation/ directory, as specified in SKILL.md.
Sanitization: No explicit sanitization, validation, or escaping of the ingested external content is defined in the instructions.- [EXTERNAL_DOWNLOADS]: The skill body contains a link to an external GitHub repository (github.com/Pimzino/spec-workflow-mcp) for schema reference purposes. This is an informational link and does not result in the execution of remote code or the download of unverifiable dependencies.

impl-logger