- Home
- Skills
- Workflow Automation
- OpenCode Delegate: Cheap Worker for Expensive Agents
OpenCode Delegate: Cheap Worker for Expensive Agents
Keep your strong model on judgement and send file edits, commands, searches and test runs to a cheap opencode worker over MCP.
$1
Works with the AI Agents you already use
OpenCode Delegate: Cheap Worker for Expensive Agents
Example session with this skill installed
/opencode-delegate Fix the three failing tests in tests/test_parser.py and add a regression test.
- Read your context and instructions
- Compiled the opencode delegate
The orchestrator writes a self-contained brief (goal, paths, constraints, acceptance command), calls delegate, reviews the worker's pytest output against the acceptance check, sends one follow-up on the same session_id for a missed edge case, then reports which files changed and the final passing test output.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop paying frontier-model prices for gruntwork. This skill turns your strong agent (Claude Opus, GPT, etc.) into an orchestrator: it keeps the judgement calls — understanding the request, planning, reviewing — and hands every file read, edit, search, command and test run to a cheap opencode worker model (DeepSeek by default, any opencode provider works).
It pairs with the free, open-source agy-opencode-mcp MCP server (MIT).
What you get
- An opt-in
/opencode-delegatemode: your agent only delegates when you ask, otherwise it works normally. - A guided setup for Claude Code, Cursor, Windsurf, VS Code Copilot and Antigravity CLI — if the server isn't installed, the agent walks you through it.
- A briefing playbook (goal, paths, constraints, acceptance check, report) so the worker gets it right the first time instead of burning retries.
- Rules for parallel fan-out with
delegate_batch, async long jobs, session follow-ups, and handling each failure state.
Why it saves money Expensive tokens stay on judgement; volume goes to the cheap model. Long file dumps and command output never enter the orchestrator's context window — only the worker's summary does.
Does not include a model subscription: the worker uses whatever provider the user has authenticated in opencode. The worker runs with opencode permissions auto-approved inside the given workspace. Without the MCP server installed the skill only guides setup.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 3 days ago
Requires
Permissions
File Scopes
The skill itself runs no scripts. The MCP server spawns `opencode run` locally; network calls go to whichever model provider the user configured in opencode.