freshcrate
Home > AI Agents > codebot-ai

codebot-ai

Safe, local-first autonomous coding agent. Policy-governed, audit-trailed, sandboxed. Works with any LLM.

Description

Safe, local-first autonomous coding agent. Policy-governed, audit-trailed, sandboxed. Works with any LLM.

README

CodeBot AI

Autonomous AI coding agent with built-in governance. Any LLM. Runs locally. MIT licensed.

npm versionlicensetests

Quick Start

npm install -g codebot-ai
codebot --setup                    # auto-detects local and cloud LLMs
codebot "refactor auth to use JWT" # run a task
codebot --dashboard                # web UI at localhost:3120
codebot --solve https://github.com/you/repo/issues/42  # autonomous fix -> PR

Hero Workflow: --solve

Point CodeBot at a GitHub issue and walk away. It delivers a tested PR with a full audit trail.

codebot --solve https://github.com/you/repo/issues/42

8-phase pipeline:

  1. Parse issue -- extract requirements from the GitHub issue
  2. Clone repo -- shallow-clone the target repository
  3. Analyze -- map the codebase and locate relevant files
  4. Install deps -- detect package manager, install dependencies
  5. Fix -- apply code changes guided by the issue description
  6. Test -- run the project's test suite, iterate until green
  7. Self-review -- audit the diff for regressions and style violations
  8. PR -- open a pull request with a structured audit trail

Key Features

  • Governance-first โ€” every tool call passes through a constitutional safety engine (CORD) that risk-scores actions across 14 dimensions before execution
  • Any LLM, anywhere โ€” 8 providers: run fully local with Ollama/LM Studio/vLLM, or connect to Anthropic, OpenAI, Google, DeepSeek, Groq, Mistral, xAI
  • 32 built-in tools โ€” code editing, shell, Chrome automation, Git, Docker, databases, web search, deep research, scheduled routines, and more
  • Cryptographic audit trail โ€” SHA-256 hash-chained logs with SARIF export for CI integration

Comparison

CodeBot AI Copilot Cursor Claude Code Devin SWE-agent
Self-hosted / local LLM Yes No No No No Yes
Any LLM provider 8 GPT Mixed Claude Proprietary Any
Safety engine + audit trail Yes No No No No No
Autonomous issue-to-PR Yes No No No Yes Yes
Free / MIT Yes $10-39/mo $20/mo $20/mo $500/mo Free

Demo

See the --solve pipeline in action: Live Demo

Architecture

User --> Agent Loop --> Tool Router --> CORD Safety Engine --> Execution
              |                              |                    |
         8 LLM Providers              14-dimension risk      32 tools
         (local + cloud)              scoring + audit      (code, shell,
                                                            browser, git...)

Extend

import { Agent, AnthropicProvider } from 'codebot-ai';

const agent = new Agent({
  provider: new AnthropicProvider({
    apiKey: process.env.ANTHROPIC_API_KEY,
    model: 'claude-sonnet-4-6',
  }),
  model: 'claude-sonnet-4-6',
  autoApprove: true,
});

for await (const event of agent.run('list all TypeScript files')) {
  if (event.type === 'text') process.stdout.write(event.text || '');
}

Custom tools via .codebot/plugins/ ยท MCP servers via .codebot/mcp.json ยท VS Code extension ยท GitHub Action


Docs ยท Changelog ยท Security ยท Contributing

MIT โ€” Ascendral

Release History

VersionChangesUrgencyDate
v2.10.0## What's New in v2.10.0 ### --solve: Autonomous Issue-to-PR Pipeline Point CodeBot at a GitHub issue, get back a reviewed PR with full audit trail. ``` codebot --solve https://github.com/you/repo/issues/42 ``` 8 phases: Parse โ†’ Clone โ†’ Analyze โ†’ Install โ†’ Fix โ†’ Test โ†’ Self-review โ†’ PR ### Electron Desktop App - Signed, notarized, Gatekeeper-approved (macOS arm64) - 113MB DMG (down from 343MB) - Auto-restart on crash with exponential backoff - Network drop recovery with reconnecting overlay - Medium3/22/2026

Dependencies & License Audit

Loading dependencies...

Similar Packages

skalesYour local AI Desktop Agent for Windows, macOS & Linux. Agent Skills (SKILL.md), autonomous coding (Codework), multi-agent teams, desktop automation, 15+ AI providers, Desktop Buddy. No Docker, no terv10.0.4
kotefAI dev that actually gets things done0.0.0
tsunamiautonomous AI agent that builds full-stack apps. local models. no cloud. no API keys. runs on your hardware.main@2026-04-21
GenericAgentSelf-evolving agent: grows skill tree from 3.3K-line seed, achieving full system control with 6x less token consumptionmain@2026-04-21
opencode-telegram-botOpenCode mobile client via Telegram: run and monitor AI coding tasks from your phone while everything runs locally on your machine. Scheduled tasks support. Can be used as lightweight OpenClaw alternav0.17.0