Open-source, self-hosted AI agent for your repos. Issue triage, PR review, auto-fix, merge queue, conflict resolution — one binary, your infrastructure, your AI keys.
Free & open source. Self-hosted. Your AI keys. Your data. Works with GitHub, GitLab, Gitea, on-premise.
Issue triage, PR review, auto-fix, merge queue and conflict resolution — all from one self-hosted Rust binary. Audit it, fork it, run it on your servers.
Automatically classify, label, and prioritize every new issue using AI. Detect duplicates, request missing info, identify quick wins. Configurable label rules, periodic retriage, custom AI prompts.
Every PR gets an instant AI summary, risk assessment, and review checklist. Line-by-line inline code review catches bugs before humans look. Auto-label by type and risk level.
Simple bugs get fixed automatically. wshm generates a draft PR, runs your test suite, scans for security issues. Gated by confidence thresholds, collaborator auth, and human review.
PRs are scored and ranked by readiness. Auto-merge above threshold. Detect and resolve merge conflicts automatically. Never force-pushes.
Real-time overview of your repo. Issues, PRs, merge queue, stats, drift analysis. All in your terminal.
wshm tui --repo rtk-ai/rtk
Same overview, in your browser. Issues, PRs, merge queue, activity, backups, settings — bundled in the OSS binary, served from the daemon.
wshm daemon → http://localhost:3000
The full agent — triage, review, auto-fix, merge queue, conflict resolution — is on GitHub under SSPL-1.0. Read the code, fork it, run it on your own infrastructure. No SaaS lock-in, no telemetry on by default, no per-seat billing.
Every prompt, every safety guard, every webhook handler is in the repo. Trust through reading, not marketing.
Runs on your servers, your AI keys, your repo tokens. Source code never leaves your infrastructure.
SSPL-1.0 lets you adapt wshm to your workflow. Custom prompts, custom rules, custom integrations — ship them in your fork.
The OSS edition is the full agent — not a crippled trial. The optional Pro tier adds a hosted dashboard, SSO and team analytics on top.
The full agent is free and open source. Pro adds a hosted dashboard, SSO and team analytics for organisations that want them.
The full agent. Self-hosted on your infrastructure.
Everything in OSS, plus hosted dashboard and team features.
Pro plus dedicated support, SLA and on-prem help.
One binary. No SaaS dependency. Runs where your code lives.
Public and private repos. Personal and organization. GitHub Action or daemon. Webhooks or polling.
GitLab.com or self-managed. CI pipeline integration. Same binary, same config.
Your servers, your network. No data leaves your infrastructure. Gitea, Forgejo, Azure DevOps Server.
wshm is a single static binary. Deploy it however you want:
systemd service, auto-update, any Linux
docker run wshm daemon, rootless Podman supported
Single-pod deployment, ConfigMap for config.toml
No server needed, runs on every event
Works with Claude Max/Pro/Team subscriptions (no API key needed).
Or use any API provider. Or run local models with Ollama.
Battle-tested security. Multiple audit rounds. Defense in depth.
--apply required. See what would happen before it happens.
XML fence escaping, confidence clamping, summary truncation, input sanitization.
Collaborator check on all slash commands. Webhook HMAC signature verification.
Conflict resolution uses new commits. Your history is always safe. Draft PRs by default.
One Rust binary, the full agent, on your servers. Read the source, run it locally, fork it for your workflow.