The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Node.js scripts to manage local model data and executes the GitHub CLI (gh) to dispatch workflows and monitor repository state.
[EXTERNAL_DOWNLOADS]: The skill interacts with GitHub's official services to trigger the update-models.yml workflow and track Pull Request status. These interactions target well-known services and are consistent with the skill's maintenance purpose.
[DATA_EXFILTRATION]: Configuration metadata is read from local JSON files and transmitted to GitHub via the update workflow. No sensitive files (such as credentials or SSH keys) are accessed or transmitted.
[PROMPT_INJECTION]: The skill processes upstream model data from ee/apps/inference/src/models/base.json and incorporates it into source code in packages/types/src/den/inference.ts. This indirect ingestion surface is mitigated by proper sanitization.
Ingestion points: ee/apps/inference/src/models/base.json (upstream model definitions)
Boundary markers: No specific boundary markers are used in the JSON source files.
Capability inventory: File system write access to local configuration and source files; GitHub CLI execution for workflow management.
Sanitization: The skill uses JSON.stringify to escape string data before writing to TypeScript files and validates that usage factors are finite positive numbers.

openwork-models