Ogcode CLI
Self-hosted AI coding agent with a web interface: project map, decision memory, local models via Ollama and permission modes
- 148 ★ on the repository
- Last commit
- AGPL-3.0
Extended access
Why this level
- The agent runs commands and edits files, and Yolo mode removes confirmations entirely
- The README install script is piped into `sh`, so read it before running
Install
Manual install
brew tap prasenjeet-symon/tap && brew install ogcodeCommand from the README for macOS and Linux. `go install github.com/prasenjeet-symon/ogcode@latest`, winget and a Docker image are also offered.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent runs commands and edits files, and Yolo mode removes confirmations entirely
- The README install script is piped into `sh`, so read it before running
This is third-party code. Review the repository files before installing.
What it does
Ogcode is an open-source AI coding agent that you run yourself and that opens a web interface at `http://localhost:9595`. The agent builds a map of the project before reading files, outlines files with line ranges and reads only what matters. It runs builds and tests, investigates failures and edits several files at once. The permission mode is chosen per session: Ask confirms every step, Auto gates dangerous actions and Yolo works without prompts. Decision memory lets it recall what was decided earlier without replaying the whole transcript. With a local model through Ollama no API key is needed. In non-interactive mode, `ogcode run "..."` executes one prompt and prints the result, and `--output-format json` gives a machine-readable answer. The `ogcode index` command builds a semantic index of the project. The repository holds a `.agents/skills` folder with a set of skills.
Who it is for. Developers who want their own coding agent with a web interface and local models, with no tie to a vendor cloud.
Good fit when
- You need a self-hosted coding agent running on your machine
- The model must be local: Ollama with no API keys
- Work goes in sessions and you need memory of past decisions
- A task has to run as a single `ogcode run` command in a script with JSON output
Not a fit when
- You need a plugin for an agent you already use: Ogcode installs as a new separate agent
- Building from source is unavailable: `CGO_ENABLED=1` is required because of the Swift tree-sitter binding
- Yolo mode is unacceptable on a work project without version control rollback
Example request
Run ogcode run "fix the failing test in the payment module" with a limit of ten turns and show the final diffLimitations
The README offers `curl | sh`, Homebrew, winget, `go install` and Docker for installation. Building from source needs `CGO_ENABLED=1`. By default `ogcode run` does not cap the number of agent turns until you set `--max-turns`. Quality depends on the connected model, and local models are weaker than cloud ones. The last commit is dated November 11, 2026. The web interface listens on port 9595, so on a shared machine close it to outside connections.
How to disable. Delete the `ogcode` binary (`brew uninstall ogcode` or by hand), stop the container if you used Docker and clear the `~/.ogcode` data folder.
FAQ
Does Ogcode need an API key?
No, if you use a local model through Ollama. Cloud models need provider keys configured separately.
What is the difference between Ask, Auto and Yolo?
Ask confirms each step, Auto gates dangerous actions and Yolo runs everything without prompts. The mode is chosen per session.
How do I run Ogcode without the interface?
With `ogcode run "task"`, and `--output-format json` turns on machine-readable output.
What does `ogcode index` do?
It builds a semantic project index the agent uses to map the project before reading files.
How do I install Ogcode?
Through Homebrew, winget, `go install` or a Docker image, or with the script from the README. Building from source needs `CGO_ENABLED=1`.
Discussion
No questions or reviews yet.
Related
CLI toolsAn open source coding agent for the terminal and desktop with build and plan modes
An OpenCode plugin that runs a team of parallel agents with separate sessions, a shared task board and messaging
Agav is a terminal coding agent that reads a repository, edits files, runs tests and switches between seven model providers.
A terminal AI coding agent built on opencode: any model with your own keys, subagent swarms, goals and output compression.
Semantic code search, cross-session memory and shell command checks for OpenCode. Runs locally on Ollama, with MCP and CLI modes
A teaching terminal coding agent in Python, built as a "nanoGPT for agents": read it, take it apart and fork it
Comparison with similar tools
Cards that share the main search query, and how they differ.
README in short
The README presents Ogcode as a self-hosted open-source coding agent. It continues with install routes (script, Homebrew, winget, `go install`, Docker), launching the web interface, working with Ollama, permission modes and the `run` and `index` commands, then a list of what the agent can do.