Verified project record

ApodexAI/FrontierAgent

FrontierAgent is an open-source agent runtime, terminal product, and evaluation suite with native ReAct and Agent Team workflows, installable in one command without Docker. It targets long-horizon research and file-based agent work that needs an integrated runtime, terminal interface, and evaluation tooling.

Project overview

It combines a stateful single-agent workflow, a parallel Agent Team coordinator mode, a terminal TUI with approval gates and revertible file mutations, and a benchmark evaluation harness in a single installable package without a hard Docker dependency.

Project type: AI Agent · Evaluation & Observability
Use cases: Search & Research
Deployment: Refer to project documentation
License: Apache-2.0

Best for

Developers and AI engineers running long-horizon research or file-based agent work who want an integrated runtime, terminal interface, and evaluation suite.
Users on macOS or Linux who want a one-command install with no preinstall and no hard Docker dependency.

Key capabilities

A single stateful agent that researches, reads files, writes deliverables, runs commands, and iterates in a task-scoped sandbox.
A coordinator maintains a task board, delegates independent work to parallel sub-agents, collects reports, and synthesizes the result.
A frontier-agent TUI with a sidebar task board, approval gates on mutating operations, checkpoints, traces, and session resume.
A subprocess runner supporting research and file-grounded benchmarks with deterministic artifact collection, concurrency, and rerun of failures.

Limitations and risks

Parallel MCP web search does not support custom region, language, or time filters, or counts above 10 per query.
Pre-built images on GHCR are private and require docker login with an authorized account; otherwise build the checkout yourself.
If you use optional local SGLang serving, note it is pinned to reviewed NVIDIA driver/CUDA tracks, and a mismatch surfaces late as opaque CUDA or Triton kernel errors.

Getting started

Requires Git, Python 3.12, uv, and configuring an OpenAI-compatible endpoint in .env. Clone with git clone https://github.com/ApodexAI/FrontierAgent.git, run uv sync --python 3.12 --extra dev, and configure .env.
Start with uv run frontier-agent --mode react --cwd /path/to/project after configuration.

Evidence and sources

GitHub project description: 🧩 FrontierAgent, our agent framework, open-sourced alongside it — native command-line TUI, ReAct and Agent Team modes, one command on macOS and Linux, no preinstall, no hard Docke…
README: FrontierAgent is an open-source agent runtime, terminal product, and evaluation suite for long-horizon research and file-based work. The `frontier-agent` TUI ships two native work…
README: - **Native Agent Team workflow.** The coordinator decomposes the request, dispatches bounded parallel assignments, receives structured reports, and can use an optional fast report…
README: - **Approval, trace, and recovery.** Mutating operations show a diff and require approval unless `--yes` is enabled. Sessions are checkpointed, every action is traced locally, `/r…
README: - **Evaluation included.** The subprocess runner supports research and file-grounded benchmarks, deterministic artifact collection, concurrency, progress inspection, and rerunning…

Source updated at: 2026-10-22T18:31:38Z

Open original GitHub project