ZhangHanDong

ZhangHanDong / agent-spec

Public

`agent-spec` is an AI-native BDD/spec verification tool for task execution.

zhanghandong.github.ioagent-spec agent ai-coding spec
18
4
100% credibility
Found Mar 08, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

agent-spec is a tool for writing structured plans that AI agents follow to code tasks, with automatic checks to verify the results match the plan.

How It Works

1
📖 Discover agent-spec

You hear about agent-spec, a simple way to guide AI helpers with clear plans so they code exactly what you want.

2
✍️ Write your task plan

Create a straightforward plan describing the goal, fixed choices, limits, and success checks, like a detailed recipe.

3
⭐ Check your plan

Review the plan automatically to make sure it's clear, precise, and ready for your AI to follow without confusion.

4
🤖 Share plan with AI

Hand the polished plan to your AI coding helper, who reads it and builds the code step by step.

5
✅ Verify the code

Run quick checks to confirm the new code matches your plan perfectly, catching any mismatches early.

6
📋 Review results

Generate a friendly summary showing what passed, what needs fixing, and why everything works.

🎉 Code is ready!

Your verified code is safe, reliable, and exactly as planned – ready to use with confidence.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is agent-spec?

agent-spec is a Rust-based, AI-native BDD/spec verification tool for task execution on agent spec github. You write structured Task Contracts defining intent, decisions, boundaries, and completion criteria in a simple DSL supporting English and Chinese, then it lints specs, renders them for agents, and verifies code changes against boundaries and explicit tests. Humans review the contract upfront; the tool machine-checks if implementations satisfy it via CLI commands like `contract`, `lifecycle`, and `guard`.

Why is it gaining traction?

Unlike agent os vs github spec kit, it enforces boundaries on git changes (staged/worktree) and binds scenarios to real tests, making agent outputs verifiable without manual reviews. The self-bootstrapping workflow—using its own specs for development—plus pre-commit hooks and agent integrations (Claude skills, Aider/Cursor configs) hooks devs tired of unverified AI code. Early adopters get deterministic pass/fail on tasks, with JSON/MD outputs for CI/PRs.

Who should use this?

Rust devs building AI agent pipelines for code gen, like teams using Claude Code to implement features against specs. Ideal for backend engineers enforcing change boundaries in monorepos or solo devs wanting pre-commit guards on task-driven workflows. Skip if you're not in agent-heavy Rust projects.

Verdict

Worth prototyping for agent spec workflows (install via crates.io, try `init` + `lifecycle`), but 1.0% credibility and 18 stars signal early-stage maturity—docs are solid via README/examples, though test coverage needs growth. Promising for ai-native bdd/spec execution in Rust; monitor for stability.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.