Gibwork Task Compiler
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Gibwork Task Compilercompile this GitHub issue into a Gibwork task contract"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Gibwork Task Compiler
Turn vague GitHub work into precise, testable, economically scoped Gibwork tasks.
Gibwork Task Compiler is a terminal-first developer tool and MCP server that transforms an issue, request, or repository change into a structured Task Contract ready for Gibwork.
Instead of asking a worker or AI agent to interpret a vague request, the compiler produces:
a clear objective
explicit requirements
testable acceptance criteria
repository and code evidence
likely files and implementation areas
a validation/test plan
constraints and exclusions
estimated effort
recommended reward range
ambiguity, confidence, and readiness scores
The goal is simple: make work easier to understand before it becomes a paid task.
Why it exists
Gibwork provides the work marketplace and developer infrastructure. The missing layer is task preparation: turning an informal request into a specification that a worker—or an AI agent—can execute and verify without repeated clarification.
Gibwork Task Compiler fills that gap from outside the browser. It can inspect public GitHub repositories, simulate execution, validate the resulting contract, preview the exact task payload, and publish only after explicit confirmation.
Related MCP server: MCP RepoBridge
Core workflow
GitHub issue / rough request
↓
Repository inspection
↓
COMPILE
↓
Task Contract
↓
SIMULATE
↓
VALIDATE
↓
PREVIEW
↓
Human confirmation
↓
GIBWORKCLI
npm install
npm run buildCompile a request:
gibwork compile \
--request "Fix the OAuth callback bug" \
--reference https://github.com/owner/project/issues/42Simulate against the real repository:
gibwork simulate \
--contract task.json \
--reference https://github.com/owner/project/issues/42 \
--output simulation.jsonValidate the contract:
gibwork validate \
--contract task.json \
--simulation simulation.jsonPreview the exact Gibwork task without publishing:
gibwork preview --contract task.jsonPublish only after explicit confirmation:
gibwork publish --contract task.json --confirmPublishing requires SOLANA_PRIVATE_KEY and never accepts a private key as a command-line argument.
MCP server
The same workflow is available to MCP-compatible AI clients such as coding agents.
Start the server:
npm run mcpAvailable tools:
compile_task— request → Task Contractinspect_repository— GitHub repository/issue → source and test evidencesimulate_task— contract → execution forecastvalidate_task— contract + evidence → readiness decisionpreview_gibwork_task— contract → exact publish payloadpublish_gibwork_task— publish a READY contract, only whenconfirmation=true
The publish tool is intentionally separated from the analysis tools and requires explicit confirmation. This prevents an AI agent from accidentally creating a paid task while still allowing the rest of the workflow to be autonomous.
Example result
🧪 GIBWORK TASK SIMULATION
REPOSITORY
owner/project · TypeScript · 420 stars · 7 open issues
EXECUTION PLAN
1. Reproduce the reported OAuth callback failure
2. Inspect callback and session handling
3. Implement the smallest compatible fix
4. Add regression coverage
5. Run the relevant test suite
LIKELY FILES
• src/auth/callback.ts
• src/auth/session.ts
LIKELY TESTS
• tests/auth/callback.test.ts
CODE EVIDENCE
• src/auth/callback.ts: authentication/session logic detected
• src/auth/callback.ts: explicit error handling detected
• tests/auth/callback.test.ts: assertion-based test coverage detected
EFFORT
1.5–2.5 hours (likely 2h)
RECOMMENDED REWARD
USDC 60–100 (recommended 80)
CONFIDENCE
91%
RECOMMENDATION
PUBLISHDesign principles
Human approval first: compiling, simulating, and validating never publish automatically.
Testable work: important requirements need observable acceptance evidence.
Repository-aware: real source and test content is used when repository context is available.
Economic realism: effort and reward are estimated together.
Machine-readable: Task Contracts are structured for CLI automation and MCP clients.
Gibwork-native: the project uses the Gibwork SDK rather than scraping the marketplace.
Outside the browser: terminal and MCP are first-class interfaces.
Safe mutation boundary: publication is isolated behind explicit confirmation.
Roadmap
Repository renamed and project direction established
Gibwork SDK foundation
gibwork compilerequest → Task ContractGitHub issue/repository context ingestion
Requirement and acceptance-criteria extraction
Task ambiguity/readiness scoring
Effort and reward simulation
Repository source/test evidence inspection
gibwork validatecontract validationgibwork simulateworker execution forecastHuman-approved Gibwork publishing
Native MCP server
End-to-end live task demonstration
LLM reasoning adapter for deeper repository analysis
Environment
Node.js 22+ is required.
For GitHub API access to public repositories, no token is required for basic usage. A GITHUB_TOKEN can be supplied when higher API limits or private-repository support is appropriate.
For publishing:
SOLANA_PRIVATE_KEY=...
GIBWORK_ENVIRONMENT=stageUse a dedicated development wallet while testing. Never commit a private key or place one in shell scripts.
Hackathon
Built for the Gibwork Developer Hackathon: a developer-tool use case that turns unstructured work requests into precise, testable Gibwork tasks using the SDK, CLI, and MCP, entirely outside the browser.
License
MIT
This server cannot be deployed
Maintenance
Related MCP Connectors
AI-native git hosting — repos, PRs, issues, CI gates, and AI code review over MCP (60 tools).
Control plane for autonomous software labor. Agents claim objectives over MCP with audit trail.
- DazbenchOAuthapp.dazbench
Task management your AI agents can actually run. One line becomes a context-ready task over MCP.
Build, validate, deploy — HTTP APIs, cron jobs, webhooks and MCP tools — from your AI client.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI-powered GitHub interactions including repository analysis, code search, PR reviews, and more through the MCP protocol.4MIT
- AlicenseNot gradedqualityCmaintenanceEnables safe repository inspection and Docker-sandboxed command execution via MCP, with optional Codex handoff for implementation tasks.1Apache 2.0
- AlicenseCqualityBmaintenanceA policy-aware MCP server for GitHub and GitHub Actions that enables safe AI-assisted infrastructure workflows—inspecting repositories, preparing branches and pull requests, and constrained remote mutations behind explicit preview-bound approval tokens.18MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI coding agents and hosts to enforce deterministic repository boundaries via MCP, providing structured reads, supervised edits, snapshots, audits, and recovery with machine-readable evidence.MIT