- Home
- Skills
- Agents & Orchestration
- opencode delegation
Works with the AI Agents you already use
opencode delegation
Delegate coding, PR reviews, and research tasks to DeepSeek agents from Claude Code to reduce token costs.
$7.50
opencode delegation
Example session with this skill installed
Use deepseek-coder to implement the error handling in logger.ts and run the unit tests. Max 20 minute timeout.
- Read your context and instructions
- Compiled the opencode delegation
I have dispatched the task to the deepseek-coder agent. It is currently working in the local workspace with allow_edits and allow_bash enabled.
Delegation Report: logger.ts Error Handling
| Step | Action | Status |
|---|---|---|
| Dispatch | delegate_task session oc_8291_xp | COMPLETED |
| Execution | Implementation of Try/Catch blocks in Logger.log() | COMPLETED |
| Verification | npm test tests/logger.test.ts | PASSED |
Changes Summary
The agent successfully modified src/utils/logger.ts to prevent process crashes on write-stream failures. It added a fallback to console.error when the primary log file is unwritable.
// src/utils/logger.ts modified by deepseek-coder
export class Logger {
async log(message: string): Promise<void> {
try {
await this.writeToDisk(message);
} catch (error) {
console.error('Logging failure, falling back to console:', error);
// Fallback logic implemented
}
}
}
Internal Verification
I have verified the uncommitted diff. The agent claimed all tests passed; I then executed npm test tests/logger.test.ts locally and confirmed the 4 passing cases.
Next steps
- Review the uncommitted changes in
src/utils/logger.tsusinggit diff. - Run
git commit -m "feat: add error handling to logger"if the implementation meets style guides. - Close the session
oc_8291_xpor send a follow-up task.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop burning through Claude 3.5 Sonnet tokens for routine coding tasks. This skill connects Claude Code to cheaper DeepSeek models via the claude-opencode-mcp server, delegating volume work while keeping Claude as the orchestrator.
What it does
- Multi-model delegation dispatches implementation, research, and testing tasks to specific DeepSeek agents.
- Automated setup handles the installation and registration of the MCP server and global dependencies.
- Context-aware sessions maintains agent memory across multiple turns to reduce briefing overhead.
- Permission control manages read/write access and shell execution per agent type and task.
- Result verification provides a structured workflow for auditing subagent claims and diffs before committing.
How it works
- Run the
setup.mjsscript to verify the MCP bridge and globalopencode-aiinstallation. - Select an specialized agent based on the task, such as
deepseek-coderfor fixes ordeepseek-reviewerfor PR audits. - Dispatch a task using
delegate_taskwith a specific timeout, workspace path, and formatted brief. - Verify the agent's work by inspecting uncommitted diffs and running local test commands.
Frameworks & tools
Uses the Model Context Protocol (MCP), claude-opencode-mcp, opencode-ai, and DeepSeek models. Requires node and npm for server management.
Why this beats prompting it yourself
Manual model-switching loses workspace context and requires constant copy-pasting of diffs. This skill automates the handoff within your current working tree, ensuring subagents have direct file access while enforcing timeouts and specific report formats you would otherwise have to repeat.
Use cases
- Delegate routine bug fixes and feature implementations to
deepseek-coder. - Perform high-volume PR reviews and code exploration without exhausting Claude limits.
- Run autonomous diagnostic loops for failing test suites using
deepseek-tester. - Brief a second model to provide a "blind" peer review of code written by Claude.
Known limitations
Subagents cannot perform git commits or pushes. Agents lack access to files outside the current workspace. Configuration changes require a Claude Code restart to take effect.
No git commit/push access. No access to files outside the workspace. Requires manual login via terminal for API keys. Claude Code must restart to load MCP server changes.
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 16 days ago
Requires