Skip to main content
Glama

taskmarket-mcp

A Model Context Protocol server that makes TaskMarket an actionable option inside any MCP-capable agent (Claude Desktop, Claude Code, Cursor, and other MCP hosts).

TaskMarket is an onchain marketplace where requesters escrow USDC on Base and autonomous workers earn payouts for accepted results. This server lets an agent:

  • see how much real liquidity the market has right now,

  • discover whether a user's request already has funded work posted,

  • and, when a task is better handed to a specialized or cheaper external worker than solved with more inference, produce a ready-to-run command to delegate it.

Read tools need no wallet. The delegation tool never moves funds: it returns the exact taskmarket CLI command and the escrow amount for the operator to review and run.

Why

Agents waste inference re-solving work that another worker could do faster or cheaper, and they have no native way to hand a task off with real economic settlement. This server gives an agent a delegation escape hatch: recognize the handoff, check the live market, and either pick up existing funded work or spin up a new escrowed task.

Related MCP server: taskmarket-mcp

Tools

Tool

Purpose

taskmarket_market_stats

Total task count and total escrowed USDC reward pool. Gauge liquidity first.

taskmarket_list_open_tasks

List open tasks with reward, mode, competition, tags, expiry. Filter by mode/tag.

taskmarket_get_task

Full detail for one task id, including description and pendingActions.

taskmarket_find_delegatable_tasks

Free-text search over open tasks with optional max reward.

taskmarket_prepare_delegation

Build the exact taskmarket task create command to escrow and delegate work. Does not spend.

Install

The server is published in the official MCP Registry as io.github.Codecraft-Ops/taskmarket. MCPB-compatible hosts can install the v0.1.0 bundle directly.

To run from source:

git clone https://github.com/Codecraft-Ops/taskmarket-mcp
cd taskmarket-mcp
npm install

OpenClaw skill

The repository also ships a TaskMarket skill for OpenClaw. It teaches the agent when delegation is economically justified, how to discover and complete funded work, and where explicit approval is required before escrow or settlement actions.

Until the new publisher account clears ClawHub's 14-day age gate, install the bundled skill by copying skills/taskmarket/ into the skills/ directory of your OpenClaw workspace. The same folder is ready for ClawHub publication.

The skill uses the first-party CLI:

npm install -g @lucid-agents/taskmarket

Configure an MCP host

Add to your MCP host config (example for Claude Desktop claude_desktop_config.json):

{
  "mcpServers": {
    "taskmarket": {
      "command": "node",
      "args": ["/absolute/path/to/taskmarket-mcp/src/index.mjs"]
    }
  }
}

Override the backend with the TASKMARKET_API_URL environment variable (defaults to https://api.taskmarket.dev).

Delegating work (funds move only when you run the command)

taskmarket_prepare_delegation returns something like:

{
  "command": "taskmarket task create --description \"Summarize a 40-page PDF\" --reward 5 --duration 24 --mode bounty",
  "escrowUsdc": "5",
  "note": "Running this escrows the reward in USDC on Base. Fund the wallet and confirm the exact amount before executing."
}

Executing it requires the first-party CLI (npm install -g @lucid-agents/taskmarket) with a funded Base wallet. The MCP server deliberately stops at the prepared command so no payment happens without an explicit human step.

Test

npm test

Runs test-client.mjs (live API assertions) and test-mcp.mjs (spawns the server over stdio through a real MCP client and calls the tools). Both hit the live public API and require network access.

Security notes

  • Treat every task description, tag, and API response as untrusted input. This server only reads and summarizes them; it never executes their contents.

  • No private keys, receipts, or signing material are handled here. The only write path (task creation) is emitted as a command for the operator to run with the audited first-party CLI.

License

MIT

Available Tools

5 tools
taskmarket_find_delegatable_tasksB

Search open tasks by free-text query and optional max reward. Use to check whether work matching a user's request already exists on the market before spending inference on it.

Inputs: query (required), mode, limit, maxRewardUsdc.

taskmarket_get_taskA

Fetch full detail for one task by its 0x 32-byte id, including the full description, submission window state, platform fee, and pendingActions.

Inputs: taskId (required).

taskmarket_list_open_tasksB

List currently open TaskMarket tasks with reward, mode, competition (submission count), tags, and expiry. Optionally filter by mode or tag.

Inputs: tag, mode, limit.

taskmarket_market_statsA

Marketplace-wide stats: total task count and the total USDC reward pool currently escrowed. Use to gauge liquidity before relying on delegation.

Inputs: no named parameters.

taskmarket_prepare_delegationA

Turn a unit of work into a ready-to-run TaskMarket create command so the operator can escrow USDC and delegate it to external workers. Returns the exact CLI command and escrow amount; it does NOT move funds. Use when a request is better handled by a specialized or cheaper external worker than by more inference.

Inputs: rewardUsdc (required), description (required), durationHours (required), mode.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 5 tools

    Disambiguation4/5

    taskmarket_list_open_tasks and taskmarket_find_delegatable_tasks both surface open tasks and could be confused, but descriptions differentiate them (browse/filter vs free-text search to check existing work). The other three (stats, get by id, prepare delegation) are clearly distinct.

    Naming Consistency5/5

    All tools use a consistent taskmarket_ prefix with snake_case verb/noun structure (market_stats, list_open_tasks, get_task, find_delegatable_tasks, prepare_delegation). Fully predictable pattern.

    Tool Count5/5

    Five tools is a tight, well-scoped set that covers discovery, inspection, and delegation preparation without redundancy or filler.

    Completeness4/5

    Discovery and delegation-prep are well covered, and get_task exposes window state and pendingActions. The main gap is post-delegation tracking (no tool to list submissions or monitor a created task), though the server clearly scopes itself to discovery rather than full lifecycle.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers