Your AI agent,
continuously learning.EvoClaw wraps your model, scores every conversation with a reward model, injects skills in real time, and trains via cloud LoRA — automatically. No GPU. No data team. No downtime.
Five automated steps run in the background — no manual intervention, no restarts, no service interruption.
No data team. No fine-tuning pipeline. EvoClaw handles the entire learning loop in the background.
Learns from live conversations — no synthetic datasets, no offline retraining. Continuously improving from actual deployment.
NO SYNTHETIC DATARetrieves relevant skill instructions and injects them into the system prompt each turn. Instant improvement without waiting for retraining.
INSTANT BOOSTWhen the agent fails, EvoClaw auto-generates a new skill from the failure trajectory using an LLM. Learns from its own mistakes.
SELF-IMPROVEMENTTraining offloads to Tinker cloud. Any machine with network access runs the complete system — zero infrastructure overhead.
CLOUD-NATIVEServing, scoring, and training run as decoupled coroutines. Your agent responds in real time while learning happens in the background.
NON-BLOCKINGRL (GRPO) for implicit environment signals. On-Policy Distillation for richer language supervision. One config field to switch.
GRPO + OPDEvoClaw supports both lightweight signal learning and rich natural-language supervision — choose what fits your setup.
EvoClaw is model-agnostic. Use Kimi-2.5 for maximum quality, Qwen3-4B for lightweight deployment, or any Groq/OpenAI-compatible endpoint.
Best quality, long context, strong reasoning. Recommended for production.
Fast iteration, lower API costs. Great for development and constrained budgets.
Groq, OpenAI, Anthropic, or any Tinker-supported endpoint. Plug and play.
All settings are passed as a single EvoClawConfig instance — no YAML files, no env sprawl.
| FIELD | DEFAULT | DESCRIPTION |
|---|---|---|
| loss_fn | "importance_sampling" | RL loss: importance_sampling / ppo / cispo |
| use_prm | True | Enable PRM reward scoring per turn |
| use_skills | False | Inject skills into system prompt |
| batch_size | 32 | Turns before each training step |
| lora_rank | 32 | LoRA rank. Higher = more capacity |
| enable_skill_evolution | False | Auto-generate skills from failures |
| proxy_port | 30000 | Proxy listen port (default: 8080 in EvoClaw) |
Tell us about your project and model stack — we'll send a personalized setup guide, a EvoClawConfig pre-filled for your scenario, and a starter skill bank curated for your use case.
EvoClawConfig pre-filled for your scenario — just add your API key.Receive a personalized config + skill bank · No credit card · MIT licensed
Your setup guide, config template, and starter skill bank are on their way. Check your inbox.
No spam · Unsubscribe anytime · Open source
Every conversation trains your model. EvoClaw batches turns, submits to Tinker cloud, and hot-swaps updated weights — all in the background. No GPU, no downtime, no manual steps.
Training runs entirely on Tinker cloud. Any machine with network access can run the full pipeline.
CLOUD-NATIVENew LoRA weights replace old ones automatically after each step. Zero downtime, zero restarts.
ZERO DOWNTIMESame model as MetaClaw. ~200B MoE, best reasoning and long context. $4.40/M tokens on Tinker.
SAME AS METACLAWLightweight alternative on Tinker free tier. Great for development and constrained budgets.
FREE TIERTwo learning modes — Reinforcement Learning (GRPO) or On-Policy Distillation. One config field.
DUAL MODENo subscription. Top up Tinker balance and pay only for actual training compute. Start from $5.
FROM $5tm1-.... Top up at least $5 under Billing → Add to balance.evoclaw init in terminal. Paste your Groq key (free at console.groq.com), then your Tinker key. Choose model 3 — Kimi-K2.5 for best results, or model 1 — Qwen3-4B for free tier.evoclaw start. You will see Tinker: ✅ connected. Point your OpenAI client to http://localhost:8080/v1 and chat normally. Every 32 conversations, EvoClaw auto-submits a LoRA job and hot-swaps the new weights.No GPU, no cluster, no data team. Install, configure, and start — EvoClaw handles the rest.
EvoClaw Proxy v0.2.1 — localhost:8080 — evolving!No GPU. No data team. No setup headache. Just plug EvoClaw in and watch your agent improve with every conversation.