ZAP is an open-source, terminal-based AI coding agent implemented in Rust by the zap-coding-agent project. It provides an autonomous command-line environment designed to index codebases, generate multi-file modifications, and manage development tasks directly from the terminal. During daily development workflows, ZAP acts as an interactive CLI agent with dedicated slash commands such as /provider and /model. It interfaces with upstream model APIs, Claude Code provider instances, and local or remote OpenAI-compatible endpoints without requiring third-party SaaS harnesses. The agent coordinates repository interactions using tools like code_map, edit_file, batch_edit, and spawn_agent. ZAP distinguishes itself by supporting structured plan execution tuned for Small Language Models (SLMs), using deterministic step-and-verify loops with built-in watchdogs that prevent recursive verification failures. It also integrates proactive quota and rate-limit tracking across rolling usage windows alongside explicit permission gating (such as Ask and Deny modes) to control automated file edits. ZAP is licensed as open-source software and distributes standalone native binaries across platforms including Linux x86_64 and ARM64.
Custom OpenAI-compatible provider flow — Connect to self-hosted LLMs or custom proxies via interactive multi-step endpoint and model discovery workflows. - SLM structured plan execution — Guide smaller language models through mechanical single-step plans with loop-detection watchdogs and failure cutoffs. - Sub-agent and batch file operations — Coordinate complex workspace changes using tools like code_map, edit_file, batch_edit, and spawn_agent. - Claude Code & API usage tracking — Monitor quota windows and rate limits with proactive notifications and sidebar usage indicators. - Granular permission controls — Enforce strict safety boundaries with configurable Ask and Deny modes to prevent uncontrolled file mutations. - Native Rust implementation — Delivers fast terminal performance with precompiled releases for Linux x86_64 and ARM64 systems.
Use cases
Use Case: Local SLM Code Refactoring Developers running local models via vLLM or Ollama can execute structured, step-by-step refactoring plans. ZAP enforces single-step verification with automatic escalation cutoffs if the model fails verification multiple times.
Use Case: Custom Inference Proxy Integration Teams routing model requests through enterprise gateways or private endpoints can use the /provider workflow to specify custom URLs, optional authentication keys, and retrieve active model lists dynamically.
Use Case: Multi-File Codebase Navigation and Editing Engineers working across large repositories can instruct ZAP to generate a structural code map, batch-apply edits across multiple source files, and spawn sub-agents to parallelize modular development tasks.
Who it is for
ZAP is designed for software engineers seeking a fast, Rust-based terminal coding agent that works with both cloud APIs and local Small Language Models. It is especially useful for developers requiring custom OpenAI-compatible server routing and strict command execution boundaries, but is less suited for developers wanting IDE-integrated GUI assistants.
Install guide
Installation Download a prebuilt release from GitHub or build directly from source using Cargo: bash git clone https://github.com/zap-coding-agent/zap-coding-agent.git cd zap-coding-agent cargo build --release
First Run Start the agent from your repository root: bash ./target/release/zap Use slash commands inside the terminal session to select your provider and model: text /provider /model
Example Session Configure a local OpenAI-compatible endpoint and execute a code update: text /provider Custom # Enter endpoint: http://localhost:11434/v1 # Enter API key: (leave blank for local) Inspect src/llm_client.rs with code_map and add robust timeout handling to HTTP requests.
Claude Code — Claude Code is an official terminal-based agentic coding tool developed by Anthropic. Executed via the `claude` CLI command, it provides…
Keen Code — Keen Code is an open-source, terminal-based AI coding agent built in Go by developer mochow13. It runs directly inside shell…
Junie — Junie is an autonomous terminal-based AI coding agent developed by JetBrains. Built on top of the IntelliJ IDEA engine, it…
Kolega Code — Kolega Code (kolega-code) is an open-source Python-based terminal coding agent developed by the Kolega AI team. It is built to…
FetchCoder — FetchCoder is a terminal-based AI coding assistant developed under the Fetch.ai organization. Built specifically for command-line workflows, the tool provides…
Open Codex — Open Codex is an open-source, community-maintained fork of OpenAI's Codex CLI tooling, maintained by ymichael on GitHub. It provides a…