The Agent Skills Directory

[SAFE]: The skill serves as a technical reference guide for developers working on the Open Code project.
[FRONTMATTER_RULE]: The allowed-tools field correctly restricts the skill's capabilities to necessary tools like Read, Grep, Bash, and WebFetch.
[TRUST-SCOPE-RULE]: References to the official GitHub repository github.com/opencode-ai/opencode and packages like @opencode-ai/sdk are internal to the project and treated as safe.
[DATA_EXPOSURE]: Documentation includes placeholders for API keys (e.g., sk-...) which are safe documentation practices and not actual hardcoded credentials.
[COMMAND_EXECUTION]: While it documents tool execution (e.g., bash), these are architectural descriptions and standard development workflows, not hidden or malicious commands.
[INDIRECT_PROMPT_INJECTION]: The skill describes how the agent processes data like SSE events and user prompts, but does not introduce any specific malicious injection vectors.

opencode-dev