tech-leads-club/harness-toolkit

Steers Cursor and Claude Code agents with gates → follow-up → handoff → policy.

What it solves

It provides a protective layer for AI coding agents like Cursor and Claude Code, ensuring they follow strict safety and operational rules that the agents themselves cannot bypass or disable. It prevents unauthorized destructive actions, secret leaks, and inconsistent agent behavior by intercepting events before the agent can execute them.

How it works

The toolkit acts as a gatekeeper that hooks into the editor's event stream. It evaluates every action against three tiers of rules: a non-configurable "floor" that blocks dangerous commands (like deleting files outside the project), "always-on" checks that monitor for policy tampering, and optional "rails" that you enable to enforce specific workflows like linting, testing, or custom project rules. When a rule is triggered, the harness can deny the action, ask for confirmation, or inject guidance into the agent's conversation.

Who it’s for

It is designed for developers and tech leads who use AI coding agents and need to enforce security, maintain code quality, or ensure consistent development practices across their repositories without relying on the agent's own internal configuration.

Highlights

  • Unbypassable Floor: Eight core rules that cannot be disabled, protecting against destructive commands, secret access, and unauthorized policy changes.
  • Zero Runtime Dependencies: Built for performance and reliability with no external runtime dependencies.
  • Customizable Rails: Twenty-four optional capabilities, including linting/testing enforcement, dependency checks, and shell stall detection.
  • Operator Rules: Allows you to define custom markdown-based rules that mandate specific actions or proofs before an agent can proceed.
  • Observability: Provides detailed reporting commands to audit agent decisions, gate failures, and rule enforcement history.

Related

  • Project
    Prorise-cool/Claude-Code-Multi-Agent

    Claude Code Multi‑Agent is an open‑source framework that gives Anthropic’s Claude Code project awareness and expert‑agent capabilities. It uses a Python Hook system and a locally‑run Ollama model to automatically detect the codebase type, analyse user intent, load 300+ domain‑specific “Skills” (backend, frontend, testing, security, etc.), and enforce automatic documentation (DEVELOPMENT.md, KNOWLEDGE.md, CHANGELOG.md). After cloning, installing Ollama, a model, and uv, you drop your project into the repo and Claude Code instantly knows its context, suggests appropriate tools, and can be driven with slash commands like `/backend‑specialist` or workflow commands such as `/kiro/spec`. Key points: - Zero‑config start; all configuration lives in `.claude/settings.json` and `prompts.json`. - Hooks run on session start, user prompt, tool use, etc., handling project detection, intent analysis, and doc updates. - Skills are defined by `SKILL.md` files and are auto‑discovered. - Works entirely locally (no external API), integrates with Git, and is MIT‑licensed. Ideal for developers or small teams who want a reproducible, on‑premise AI coding assistant.

  • Project
    HeyCubit/effortless

    A Claude Code plugin that automatically adjusts reasoning effort per prompt to reduce API costs while providing tools to manage chat context, cache health, and session handoffs.

  • Project
    thewaltero/mythos-router

    A local CLI tool for verifiable AI-assisted coding that uses Strict Write Discipline to ensure AI file operations match the actual filesystem state, preventing hallucinations and broken writes.

  • Project
    anza-xyz/mollusk

    Mollusk is a lightweight test harness for Solana programs that runs them in a fast, minified SVM environment without the overhead of a full validator runtime. It executes program ELF files directly through the BPF Loader, supports instruction chains, automated account builders, stateful testing contexts, compute-unit benchmarking, and replayable test fixtures.