What Hive works with today, and what is planned. One row per integration. Each row links the doc that explains it. Planned and In progress rows link the tracking issue, or the page that leads to it.
Status key:
| Status | Meaning |
|---|---|
| Supported | Wired into a running code path and covered by tests. |
| Opt-in | Works, but an operator must turn it on, or it needs a per-backend escape hatch. |
| Planned | Not usable today. The linked issue tracks the work. |
| In progress | Started, not finished. The linked tracker lists what is left. |
Tiers (T1 core, T2 confined, T3 experimental) come from
backend-support-tiers.md. The source list is
KNOWN_BACKENDS in config/backends.conf. Setup for each CLI is in
backend-setup.md. T3 backends run in
container mode by default; local mode needs an explicit opt-in.
| Logo | Integration | Status | Tier | Docs |
|---|---|---|---|---|
 | Claude Code | Supported | T1 / T2 | backend-setup.md |
 | GitHub Copilot CLI | Supported | T1 / T2 | backend-setup.md |
 | Codex CLI (OpenAI) | Supported | T1 / T2 | backend-setup.md |
 | Goose (AAIF) | Supported | T1 pod / T3 local | goose-at-scale.md |
 | Google Antigravity (agy) | Opt-in | T3 | backend-setup.md |
 | Gemini CLI (hub-managed only) | Supported | Go-only, no local tier | backend-support-tiers.md |
 | OpenAI-compatible gateways: vLLM, llm-d, LiteLLM, watsonx, OpenRouter | Supported | Gateway path | inference-backends.md, gateway-path tier |
 | Muse Code | Supported | T2 | backend-support-tiers.md |
 | OpenCode | Supported | T2 (deny-list) | backend-setup.md |
 | IBM Bob | Opt-in | T3 | backend-setup.md |
 | Pi | Opt-in | T3 | backend-setup.md |
 | Aider | Opt-in | T3 | backend-setup.md |
 | Kilo Code | Opt-in | T3 | backend-setup.md |
 | Oh My Pi (OMP) | Opt-in | T3 | agent-configuration.md |
 | OpenHands (legacy CLI) | Opt-in | T3 | backend-support-tiers.md |
| Claude Managed Agents | Experimental | Hosted execution; private self_hosted spoke, report-only, USD 25/day ledger cap | agent-configuration.md, backend-support-tiers.md |
 | OpenAI Agents API | Planned | — | #14093 |
 | Omnigent | Opt-in | T3 | backend-support-tiers.md |
| Integration | Status | Execution and limitations | Docs |
|---|---|---|---|
| Apple MLX / MLX-LM through Pi | Opt-in | ClankeR on Apple Silicon macOS 14+, native Python 3.10+, Pi 1.4.0+; user-managed tool-capable models. MLX-LM runs on the Mac; Pi executes tools inside the contributor container with Docker Desktop host networking. No cloud fallback. Live Apple GPU inference remains unverified. | Install and use with just |
Apple MLX is a first-class inference-engine choice on /contribute, mapped to
the actual pi coding backend and mlx/<model> provider selection. It is not
a standalone coding CLI, hosted gateway, identity provider, or Apple endorsement.
KNOWN_BACKENDS correctly continues to list coding executors, not inference
frameworks.
The governor reads work through one provider contract. The adapters live in
src/pkg/worksource/ and are selected by governor.work_source.type.
| Logo | Integration | Status | Docs |
|---|---|---|---|
 | GitHub Issues | Supported | work-sources.md |
| GitHub Projects (v2) | Supported | work-sources.md |
 | Linear | Supported | work-sources.md |
 | Jira Cloud and Data Center | Supported | work-sources.md |
GitHub is the only forge agents can work against today. The GitLab and
Gitea/Forgejo adapters in src/pkg/forge/ cover reads and core writes
(comment, label, hold), and GitLab's can merge. Two governor writes use them,
but the GitHub-only enumeration that feeds those writes means they never fire
on a non-GitHub hive. See Forges.
| Logo | Integration | Status | Docs |
|---|---|---|---|
| GitHub.com and GitHub Enterprise | Supported | github-app-setup.md |
 | GitLab (SaaS or self-managed) | Planned | forge-app-setup.md, #14096 |
 | Gitea | Planned | forge-app-setup.md, #14096 |
 | Forgejo | Planned | forge-app-setup.md, #14096 |
The hub-side agent sandbox is off by default. See sandbox-isolation.md for the two launchers and the per-backend confinement matrix.
| Logo | Integration | Status | Docs |
|---|---|---|---|
 | Rootless Podman agent sandbox | Opt-in | sandbox-isolation.md |
 | Kubernetes Job sandbox runtime | Opt-in | sandbox-isolation.md |
 | NVIDIA OpenShell sandbox runtime | In progress (opt-in, phase 1) | sandbox-isolation.md · #17642 |
 | Agent Substrate sandbox runtime (CNCF Sandbox) | In progress (opt-in, phase 2) | sandbox-isolation.md · #17695 |
 | AAIF OpenSandbox (evaluation) | Planned | #14092 |
| Logo | Integration | Status | Docs |
|---|---|---|---|
| CNCF sandbox onboarding | In progress | cncf-sandbox-onboarding.md, #9846 |
 | GitHub Agentic Workflows (gh-aw) report-only sample | Supported (sample) | deploy/gh-aw, tiers note |
| Goose-at-scale guide upstreamed to AAIF | Planned | goose-at-scale.md, #14095 |
- Landscape and positioning — how Hive compares to nearby tools.
- Work-source provider contract.
- Forges: GitLab, Gitea and Forgejo.
- Logo sources and licenses.
Logos are trademarks of their respective owners. They are used here for identification only and do not imply endorsement.