pantheon-ai/opencode-toolkit
Complete toolkit for configuring and extending OpenCode: agent creation, custom slash commands, configuration management, plugin development, and SDK usage.
98
Quality
98%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Securityby
Advisory
Suggest reviewing before use
task.mddesign-agents/evals/scenario-2/
evals task.md
An existing agent called "code-reviewer" is not triggering reliably. Its current description is: "A helpful code review assistant."
Improve this agent to trigger correctly when users say: "review my code", "check this file for bugs", "find issues in", "LGTM?", "code quality".
Also, the current system prompt says: "The agent should check for bugs and provide feedback." Fix this to follow best practices.