Open source · MIT Licensed · OpenClaw Plugin

Stop babysitting
your AI agents.
Let them work as a team.

DevClaw turns every Telegram group into an autonomous dev team. Create issues, go to sleep, wake up to completed work. DEV writes code. QA reviews it. Failures loop back automatically.

75+
GitHub Stars
30+
Reddit upvotes per post
60-80%
Token savings reported
MIT
Open Source License

AI agents write great code.
They're terrible at managing pipelines.

Every task involves 10+ coordinated steps. Agents follow them imperfectly. The more steps, the more things break. You end up babysitting the thing you built to avoid babysitting.

💀

Forgotten steps

Label transitioned but state didn't update. Session spawned but audit log didn't write. Now you have inconsistent state and an agent trying to figure out what went wrong — which it will do poorly.

🔥

Token waste

Every new sub-agent session reads the entire codebase from scratch. A huge chunk of token usage isn't even coding — it's the agent reasoning about "what should I do next."

🧟

Zombie sessions

Sessions die, workers get stuck, state drifts. Active worker with a dead session. Worker stuck for hours. Orphaned references nobody cleans up.

💸

Wrong model, wrong job

Opus for a CSS typo. Haiku for a database migration. Without automatic model selection, you're either overpaying or under-delivering on every task.

Guardrails as code,
not as prompts.

Don't make the agent responsible for process. Move orchestration into deterministic code. The agent provides intent. Tooling handles mechanics. One atomic tool call replaces a 10-step agent checklist.

orchestrator → devclaw
// The orchestrator makes one call.
// DevClaw handles everything atomically.

task_pickup({ issueId: 42, role: "dev", projectGroupId: "-100234" })

// ✓ Resolves project from state
// ✓ Validates no active worker
// ✓ Fetches issue from GitLab
// ✓ Selects model → sonnet (standard feature task)
// ✓ Detects existing session → reuse
// ✓ Transitions label: To Do → Doing
// ✓ Updates projects.json
// ✓ Writes audit log

→ { sessionAction: "send", sessionId: "a9e4d078...", model: "sonnet" }

// No step forgotten. No state corrupted. Every time.
📋 Create issue
🔧 DEV writes code
🔍 QA reviews
Pass? Close.
🔁 Fail? Loop back.
Planning
To Do
Queued for dev
Dev
Doing
DEV agent coding
Review
Testing
QA agent reviewing
Result
Done
Issue closed
Fail
To Improve
Loops back to DEV

Telegram drives the work.
GitHub tracks the results.

Agents pick up issues, write code, create PRs, and loop through review — all visible in your group chat. The issue board updates itself.

DevClaw autonomous development updates in Telegram group chat
💬

Telegram Group Chat

DEV agents complete tasks, QA flags reviews, failures loop back — all autonomous

GitHub issues board showing DevClaw managed issues with model tier labels
🐙

GitHub Issues

208 issues closed. Junior, medior, senior labels auto-assigned by complexity

Four tools replace hundreds
of lines of orchestration logic.

🎯

Per-project isolation

Each Telegram group is a separate project. Own queue, workers, session state. The moment projects share anything, you get cross-contamination. DevClaw prevents that by design.

⚛️

Atomic operations

Every operation that touches multiple things succeeds or fails as a unit. Label, state, session, audit log — all in one call. Roll back on any failure. No partial state corruption.

🔄

Session reuse

When a DEV finishes task A and task B is waiting, it goes to the existing session. The worker already knows the codebase. No 50K token re-read. Massive savings on sequential tasks.

🧠

Smart model selection

Haiku for typos. Sonnet for features. Opus for architecture. The right model for every task, selected automatically by complexity. Override anytime with modelOverride.

Token-free scheduling

A heartbeat scans queues and dispatches tasks through pure CLI calls. Zero LLM tokens burned on "what should I do next." The model only activates when there's code to write.

🏥

Self-healing health checks

Automated detection for zombies, stale state, and orphaned references. Auto-fix the straightforward cases, flag the ambiguous ones. Your pipeline recovers without you.

Sleep through the night.
Wake up to completed issues.

Model tiering, session reuse, and token-free scheduling compound to massive savings. But the real win is reliability — across multiple projects, simultaneously.

60-80%
Token savings vs. one large model with fresh context each time
30-50%
Cost reduction on simple tasks via automatic model tiering
~50K
Tokens saved per session reuse by not re-reading the codebase
0
Tokens spent on scheduling and orchestration decisions

Think in roles, not model IDs.

A CSS typo doesn't need your most expensive developer. A database migration shouldn't go to the intern.

Junior Haiku Typos, CSS fixes, renames, copy changes ~70% cheaper
Medior Sonnet Features, bug fixes, multi-file changes Balanced
Senior Opus Architecture, migrations, security, system-wide refactoring Full power

Built in the open.
Used by its creator every day.

Without atomic state management you are just funding expensive zombie sessions.

Community feedback on Reddit — validating the guardrails-as-code approach

Ship while you sleep.

DevClaw is open source, MIT licensed, and built on OpenClaw. Install it, point it at a repo, and let your agents do the work.