The Agent Skills Directory

[SAFE]: The skill consists of instructional text and configuration for managing workflow states. No executable code, remote downloads, or credential exposures were found. All referenced skills (e.g., and-workflow-contract, and-pack) appear to be part of the same vendor's toolset.
[PROMPT_INJECTION]: The skill uses natural instructional language and does not attempt to bypass safety filters or override agent behavior. The process involves reading external delivery units and GitHub issues, which creates a surface for potential indirect prompt injection. However, the skill focuses on status management and lacks high-privilege execution capabilities that would make such an attack impactful.
Ingestion points: External delivery units and GitHub issues processed via the Read Delivery Unit step.
Boundary markers: None explicitly defined for differentiating untrusted external content from instructions.
Capability inventory: Capabilities are limited to recording ownership status and appending receipt text to GitHub issues.
Sanitization: No specific sanitization or filtering is defined for the content read from external delivery units.

and-claim