finds.dev← search

// the find

darrenhinde/OpenAgentsControl

★ 4,315 · TypeScript · MIT · updated Mar 2026

AI agent framework for plan-first development workflows with approval-based execution. Multi-language support (TypeScript, Python, Go, Rust) with automatic testing, code review, and validation built for OpenCode

OpenAgentsControl is a workflow layer on top of OpenCode that enforces a plan-then-approve-then-execute cycle for AI coding agents. The core idea is that you define your project's coding patterns once in markdown files, and agents load those patterns before generating code, so the output matches your conventions instead of generic defaults. Aimed at teams with established codebases who want to stop refactoring AI-generated code.

The pattern-as-files approach is genuinely useful — storing your conventions in versioned markdown that agents load is simpler and more auditable than prompt injection hacks. Approval gates before any file write or bash execution are a real safety property, not just a marketing claim. The MVI (Minimal Viable Information) principle of keeping context files under 200 lines and lazy-loading only what's needed is a thoughtful answer to the context-bloat problem that kills most agent frameworks. The subagent delegation model (ContextScout, TaskManager, CoderAgent, TestEngineer as distinct roles) maps well to how you'd actually split work on a real team.

The repo is essentially a collection of markdown prompt files with an install script — there's almost no executable code to audit, which makes the '80% token reduction' and other performance claims impossible to verify. It's entirely dependent on OpenCode as the runtime, so you're one breaking OpenCode release away from nothing working, and the project board links in the README 404. The multi-language support claim (TypeScript, Python, Go, Rust, C#) comes down to whether your context files describe those patterns well enough, not any actual language-specific tooling — it's the user's responsibility to write those files, not the framework's. 4300 stars on what is largely a prompt template library raises questions about organic vs. viral growth.

View on GitHub →

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →