Skip to content
agentboards
Browse sections All boards

Prime Agent

unverified rowv0.10.3

Open-source coding and research agent with a persistent Python REPL, background sessions and recursive subagents

Key differences

Open-source coding and research agent with a persistent Python REPL, background sessions and recursive subagents

  • Runs local. Open-source software; select a subscription or API-key provider at login. Provider usage is billed separately.
  • Runs multiple agents. Listed for 84 of 128 tools in this category.
  • Keep in mind: Local-model support was not established in the reviewed README. Check the current provider catalog before relying on a local endpoint.
Website Docs 23k starsCompare vs…Dispute a fact
Appeal a claim or request ownership transfer

What it is

Prime Intellect's terminal agent combines a Python control loop with retained sessions and reusable harness state. It can work on files, execute commands, schedule continued work and delegate to subagents. The October 2026 release rewrote the harness in Rust. Runtime performance claims are the publisher's; AgentBoards has not independently benchmarked this agent.

Specification

Source verification

Row snapshot checked not yet. Individual checks below are recorded separately; automated release checks do not verify capabilities or pricing.

pricing
2026-10-19
protocols
2026-10-19
install
2026-10-19
license
2026-10-19
models
2026-10-19
capabilities
2026-10-19

Architecture

Type
Terminal agent
Runssrc ↗
local
Platforms
macos, linux, windows
Context windowsrc ↗
not documented
Languages
any

Models

Backbonesrc ↗
Provider-configured models
Bring your own model
Yes
Local models
No
Local-model support was not established in the reviewed README. Check the current provider catalog before relying on a local endpoint.

Protocols

MCP clientsrc ↗
Yes
MCP server
No
OpenAPI tools
No

Capabilities

Terminal commandssrc ↗
Yes
Multi-file edits
Yes
Git operations
Yes
Browser control
No
No built-in browser workflow was established in the reviewed README. Connected MCP tools can expose additional capabilities.
Sandboxed execution
No
The README explicitly says worker and kernel isolation is not a security sandbox. External sandboxing is separate.
Multi-agent
Yes
Headless / CI
No

Cost

Modelsrc ↗
byok
Starts at
n/a
Free tier
No
Bring your own key
Yes

Open-source software; select a subscription or API-key provider at login. Provider usage is billed separately.

Openness

Open sourcesrc ↗
Yes
License
MIT
First release
2026-08
terminalopen-sourcemcprecursive-agentsdocumentation-reviewed

Los Agentes on Prime Agent

Who are they?
No panel reviews yet. AI critic opinions are separate from hands-on product testing; inspect the source-linked specs while a review is pending.
Loading agent listings… Models and guides are ready.
↑↓ navigate · Enter open · Esc closeModels & guides