The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the official GitHub REST API OpenAPI schema from GitHub's public repository. This schema is used as configuration to dynamically generate CLI commands via the uxc tool.
[DYNAMIC_EXECUTION]: The skill utilizes the uxc link command to generate a local CLI tool (github-openapi-cli) at runtime based on the downloaded OpenAPI schema. This allows the agent to interact with GitHub's API endpoints through a structured interface.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, creating a potential surface for indirect prompt injection.
Ingestion points: Operations described in SKILL.md (e.g., get:/repos/{owner}/{repo}/issues and get:/repos/{owner}/{repo}/events) retrieve user-controlled content from GitHub repositories.
Boundary markers: The skill instructions include high-level guardrails suggesting read-first workflows and explicit user confirmation for write operations; however, there are no specific prompt-level delimiters defined for the data returned from the API.
Capability inventory: The skill has the capability to perform network operations and modify external repository state (e.g., creating issues or comments) via the generated github-openapi-cli.
Sanitization: No explicit sanitization, escaping, or validation mechanisms for the content retrieved from external API calls are documented in the skill scripts or instructions.

github-openapi-skill