Compose a PRView SourceOpen Issue

Supported agents & inference engines

Hive supports agent CLIs, inference engines, and OpenAI-compatible model gateways by backend ID. These are supported software integrations; inclusion does not imply endorsement or a partnership unless marked Partner. TypeSafe AI is a Hive Commons partner.

Claude Code by Anthropic Claude Code

  • Owner: Anthropic
  • What Hive does: launches Claude Code as an agentic CLI, applies Hive’s prompt, permissions, audit, and queue controls, and records the configured model with the agent.
  • Hive config: backend: claude
  • Official link: Claude Code

GitHub Copilot CLI by GitHub GitHub Copilot CLI

  • Owner: GitHub
  • What Hive does: launches Copilot CLI for unattended agent work with Hive-managed repository scope, permissions, and audit trails.
  • Hive config: backend: copilot
  • Official link: GitHub Copilot CLI reference

Goose by Block Goose

  • Owner: Block
  • What Hive does: launches Goose as an agent CLI and delivers Hive prompts and kicks through the Goose runtime.
  • Hive config: backend: goose
  • Official link: Goose

OpenAI Codex CLI by OpenAI OpenAI Codex CLI

  • Owner: OpenAI
  • What Hive does: launches Codex CLI for agent tasks and passes Hive’s selected model and reasoning-effort settings where supported.
  • Hive config: backend: codex
  • Official link: OpenAI Codex CLI

Pi by Earendil Works Pi

  • Owner: Earendil Works
  • What Hive does: launches the Pi coding-agent CLI as a long-running interactive backend and sends Hive kicks when the CLI is ready.
  • Hive config: backend: pi
  • Official link: Pi

IBM Bob by IBM IBM Bob

  • Owner: IBM
  • What Hive does: launches the IBM Bob / bobshell CLI with Hive-managed headless authentication, approval, and workspace trust settings.
  • Hive config: backend: bob
  • Official link: IBM watsonx Code Assistant

Aider open source coding agent Aider

  • Owner: Aider open source project
  • What Hive does: launches Aider as an agent CLI and can apply Hive’s per-agent prompts, repository context, and audit controls.
  • Hive config: backend: aider
  • Official link: Aider

Gemini CLI by Google Gemini CLI

  • Owner: Google
  • What Hive does: launches Gemini CLI for operators whose Google access still supports that CLI and treats it as an agentic backend.
  • Hive config: backend: gemini
  • Official link: Gemini CLI

Google Antigravity CLI by Google Google Antigravity CLI

  • Owner: Google
  • What Hive does: launches the Antigravity CLI (agy) as Google’s current Hive-supported agent CLI, with Hive managing unattended mode and reasoning effort.
  • Hive config: backend: agy
  • Official link: Google Antigravity CLI

OpenCode agent CLI OpenCode

  • Owner: OpenCode project
  • What Hive does: launches OpenCode as a supported terminal coding-agent backend and routes Hive prompts to it.
  • Hive config: backend: opencode
  • Official link: OpenCode docs

Kilo Code agent CLI Kilo Code

  • Owner: Kilo
  • What Hive does: launches Kilo Code as a supported agentic CLI backend for Hive-managed work.
  • Hive config: backend: kilo
  • Official link: Kilo CLI

Muse Code by Meta Muse Code

  • Owner: Meta
  • What Hive does: launches Muse Code as a supported CLI backend when installed by the operator.
  • Hive config: backend: muse
  • Official link: Muse Code docs

Oh My Pi

  • Owner: can1357
  • What Hive does: launches Oh My Pi through the omp backend and passes Hive’s selected model and approval behavior.
  • Hive config: backend: omp
  • Official link: Oh My Pi

Flue

  • Owner: Flue / withastro
  • What Hive does: runs a report-only external-execution binding pilot through pkg/extwork and the Flue adapter. The binding admits, observes, and verifies external workflow receipts without publication credentials; it is linked when Hive is built with the extwork_flue build tag.
  • Hive config: runs.external.flue (pilot, default off)
  • Official link: Flue; Hive Flue example; external workflow admission design

vLLM inference engine vLLM

  • Owner: vLLM project
  • What Hive does: routes inference through a vLLM-compatible OpenAI API endpoint while Hive drives agent turns and audit metadata.
  • Hive config: backend: vllm
  • Official link: vLLM

llm-d inference engine llm-d

  • Owner: llm-d project
  • What Hive does: routes inference through an llm-d OpenAI-compatible endpoint for self-hosted model serving.
  • Hive config: backend: llm-d
  • Official link: llm-d

LiteLLM by BerriAI LiteLLM

  • Owner: BerriAI
  • What Hive does: routes model calls through LiteLLM as an OpenAI-compatible proxy for centrally managed models and keys.
  • Hive config: backend: litellm
  • Official link: LiteLLM

IBM watsonx.ai by IBM IBM watsonx.ai

  • Owner: IBM
  • What Hive does: routes model calls through watsonx.ai’s OpenAI-compatible gateway and handles the IBM project and token settings required by Hive configuration.
  • Hive config: backend: watsonx
  • Official link: IBM watsonx.ai

OpenRouter model gateway OpenRouter

  • Owner: OpenRouter
  • What Hive does: uses OpenRouter as a named model gateway for OpenAI-compatible routing when configured by an operator.
  • Hive config: backend: openrouter
  • Official link: OpenRouter

Jev by TypeSafe AI Jev

  • Owner: TypeSafe AI — Partner
  • What Hive does: in Hive v6, uses Jev as an optional advisory smart classifier (classifier.backend: jev) that measures disagreement with deterministic keyword rules and suggests rules for approval in Settings → Smart classifier. It does not change routing. Jev returns typed choice, score, and probability answers with confidence, and bills input tokens. Per-agent jev_mode is planned.
  • Hive config: classifier.backend: jev with OpenRouter connected or JEV_API_KEY for TypeSafe AI. OpenRouter model ID: typesafe/jev-1.13.
  • Official link: Jev System docs; TypeSafe AI API; TypeSafe AI

TypeSafe AI is a Hive Commons partner.

Anthropic model provider Anthropic

  • Owner: Anthropic
  • What Hive does: records and routes Anthropic-backed models through the configured CLI or gateway while preserving Hive audit metadata.
  • Hive config: backend: anthropic
  • Official link: Anthropic

OpenAI model provider OpenAI

  • Owner: OpenAI
  • What Hive does: records and routes OpenAI-backed models through the configured CLI or gateway while preserving Hive audit metadata.
  • Hive config: backend: openai
  • Official link: OpenAI

DeepSeek model provider DeepSeek

  • Owner: DeepSeek
  • What Hive does: routes DeepSeek-backed models through a configured OpenAI-compatible gateway such as LiteLLM.
  • Hive config: backend: deepseek
  • Official link: DeepSeek

Crustify / Wavefront

  • Owner: Crustify / Wavefront projects
  • What Hive does: adds the Wavefront work source to list ready nodes from C/C++→Rust migration graphs as run-stage work items. Wavefront remains authoritative for graph dependency closure, cycles, waves, and batching; Hive records receipts when nodes complete and leaves publication out of scope.
  • Hive config: governor.work_source.wavefront (default disabled)
  • Official link: Crustify; Wavefront; Hive work-source docs

Infrastructure thanks

Akamai (Linode) infrastructure provider Oracle Cloud (OKE) infrastructure provider Cloudflare DNS edge and tunnels provider GitHub Copilot AI inference supporter Bluehost domain hosting provider

Hive Commons thanks Akamai (Linode) and Oracle Cloud (OKE) for Kubernetes infrastructure donated through the CNCF, Cloudflare for DNS, edge and tunnels, GitHub Copilot for AI inference donated to CNCF projects through the CNCF, and Bluehost for domain hosting.

Source control, work sources, and sign-in

The product landing page groups runtime integrations separately from the agent CLI and inference list above. Current shipped surfaces are:

  • Source control: GitHub and GitHub Enterprise are the production GitHub App path. The forge-neutral adapter layer includes GitLab and Gitea / Forgejo issue and change-request flows as those paths graduate.
  • Work sources: GitHub Issues (default), GitHub Projects, Linear, Jira Cloud, Jira Data Center / Server (deployment: datacenter), Spektacular run stages, and Crustify / Wavefront migration graphs.
  • Sign-in: Microsoft, GitHub, IBM, Google, and other OIDC providers by hub configuration.

Jira Data Center / Server support uses Jira REST API v2, preserves context-path base URLs, and supports Personal Access Token bearer auth or basic auth fallback. Custom CA and TLS trust settings are still in progress and are not listed as shipped stable functionality.

Recent Hive platform additions

  • Swarm mode: per-repository swarms with idle-unlock, themes, Discord announcements, and /api/leaderboard/swarm.
  • The Commons: contributors can subscribe to multiple hives, rank them, and choose ranked, spread, or neediest routing; hivectl hives web opens a local management UI.
  • Teams and achievements: distro / OS / agent team leaderboards are exposed at /api/leaderboard/teams, and Achievements 2.0 adds Solo, Dual, Fireteam, and Raid tiers plus the local-model track.
  • Edge-only v6: admin MCP (/api/admin/mcp, cmd/hive-admin-mcp) and the dashboard Extensions tab for Spektacular are on the v6 edge branch until that line is promoted.