Rig
tryrig.ai- Category
- AI Agents
- Rank
- No. 2348Tools index
Previous survey · No. 2268 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Coding
- Interfaces
- CLI
- Date
About
Rig is a single-binary command-line agent runtime that connects to any OpenAI-API-compatible model, local (llama.cpp, Ollama) or hosted (OpenRouter), and runs an agent loop that reads, edits, and tests code in a repository. It supports sandboxed background workers, cron-style scheduled jobs, custom Python-based tools, and JSON configuration files for models, worker fleets, themes, and standing instructions, and is released under the MIT license.
What it does
The site says Rig is a terminal coding agent shipped as one binary that you point at a model endpoint of your choosing. It describes an agent loop of a few hundred lines that reads a repository, edits files, runs tests and records what it learned, with plugins, workers and scheduled jobs around it.
Stated on the product site
- License
- MIT
- Platforms
- Linux and macOS, on amd64 and arm64
- Model endpoints
- Any endpoint that speaks the OpenAI API, local or hosted, including Ollama and OpenRouter
- Built-in tools
- Thirteen, including bash, edit, delegate and scheduler
- Plugins
- One Python file per tool, held as pending until the user approves it
- Installation
- Shell installer, checksum-verified, updates itself in place
Not stated on the site
- The page does not show whether Rig has any paid tier or hosted service.
- The token, cache and session figures come from the project's own state store; the page does not show how they were collected beyond that.
- The page cites a git log of 1283 commits but does not link where that log can be read.
- How well the agent works with small local models is asserted on the page rather than shown by an outside measurement.
Written from the product site at tryrig.ai.
Tags
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.