Large language model (LLM) agents are increasingly used for complex tasks, yet deployed agents often remain static, failing to adapt as user needs evolve. This creates a tension between the need for continuous service and the necessity of updating capabilities to match shifting task distributions. On platforms like OpenClaw, which handle diverse workloads across 20+ channels, existing methods either store raw trajectories without distilling knowledge, maintain static skill libraries, or require disruptive downtime for retraining. We present MetaClaw, a continual meta-learning framework that jointly evolves a base LLM policy and a library of reusable behavioral skills. MetaClaw employs two complementary mechanisms. Skill-driven fast adaptation analyzes failure trajectories via an LLM evolver to synthesize new skills, enabling immediate improvement with zero downtime. Opportunistic policy optimization performs gradient-based updates via cloud LoRA fine-tuning and Reinforcement Learning with a Process Reward Model (RL-PRM). This is triggered during user-inactive windows by the Opportunistic Meta-Learning Scheduler (OMLS), which monitors system inactivity and calendar data. These mechanisms are mutually reinforcing: a refined policy generates better trajectories for skill synthesis, while richer skills provide higher-quality data for policy optimization. To prevent data contamination, a versioning mechanism separates support and query data. Built on a proxy-based architecture, MetaClaw scales to production-size LLMs without local GPUs. Experiments on MetaClaw-Bench and AutoResearchClaw show that skill-driven adaptation improves accuracy by up to 32% relative. The full pipeline advances Kimi-K2.5 accuracy from 21.4% to 40.6% and increases composite robustness by 18.3%. Code is available at https://github.com/aiming-lab/MetaClaw.
Figures & tables
Figure 1: Overview of . The framework improves the meta-model M=(θ,S) via two complementary loops operating at different timescales. Skill-driven fast adaptation (left) analyzes failed trajectories and instantly expands the skill library S without parameter updates, taking effect immediately for subsequent tasks. Opportunistic policy optimization (right) accumulates post-adaptation trajectories and, once sufficient data is available, leverages idle signals (sleep, inactivity, calendar) detected by the Opportunistic Meta-Learning Scheduler to trigger RL-based weight updates on θ via Cloud LoRA fine-tuning.
Part I (30 days, 346 Q)
Part II (14 days, 588 Q)
Model
Condition
Acc. (%)
Compl. (%)
Acc. (%)
Compl. (%)
GPT-5.2
Baseline
41.1
14.7
44.9
58.4
GPT-5.2
MetaClaw (Skills)
44.0
17.1
49.1
67.5
Kimi-K2.5
Baseline
21.4
2.0
21.1
18.2
Kimi-K2.5
MetaClaw (Skills)
28.3
2.0
26.9
33.8
Kimi-K2.5
MetaClaw (Full)
40.6
16.5
39.6
51.9
Table 1: Main results on MetaClaw-Bench Parts I and II. Acc.: mean per-question accuracy. Compl.: file-check completion rate. MetaClaw (Full) is evaluated for Kimi-K2.5 only. Best result per model per part is bolded .
Metric
Baseline
+ MetaClaw (Skills)
Relative Change
Stage retry rate ( ↓ )
10.5%
7.9%
↓ 24.8%
Refine cycle count ( ↓ )
2.0
1.2
↓ 40.0%
Pipeline stage completion ( ↑ )
18 / 19
19 / 19
↑ 5.3%
Composite robustness score ( ↑ )
0.714
0.845
↑ 18.3%
Table 2: MetaClaw (Skills-Only) on AutoResearchClaw, a 23-stage autonomous research pipeline. Skill injection alone yields consistent improvements across all robustness metrics without requiring RL weight updates.
Figure 4
Case 1 MetaClaw (Skills)
Case 2 MetaClaw (Full)
Model
GPT-5.2
Kimi-K2.5
Day / Round
Day 19 / Round 4
Day 18 / Round 6
Task type
File-check
File-check
Task instruction
Update sprint8_board.json : set T-405/T-406 to "done" , add completed_at fields.
Append a deployment record to deploy_log.json with fields timestamp (ISO 8601+TZ), env , status , and changes .
Baseline response (score: 0)
Table 3: Representative case studies. Case 1 shows skill-driven fast adaptation (MetaClaw Skills, GPT-5.2); Case 2 shows the full pipeline (MetaClaw (Full), Kimi-K2.5). Both recover from score 0 to score 1.0; the mechanisms differ fundamentally.
Every output file must include created_at , author , status
Day 06
P4
Backup
Create <file>.bak before modifying any existing file
Day 08
P5
Completion log
Append [DONE] <timestamp> | <task_id> | <summary> to done.log
Day 10
Appendix
Table 4: The five implicit preference rules in MetaClaw-Bench Part II, introduced progressively across 6 learning arcs. Each rule is verified by a dedicated automated checker script.
Large language model (LLM) agents such as OpenClaw rely on reusable skills to perform complex tasks, yet these skills remain largely static after deployment. As a result, similar workflows, tool usage patterns, and failure modes are repeatedly rediscovered across users, preventing the system from improving with experience. While interactions from different users provide complementary signals about when a skill works or fails, existing systems lack a mechanism to convert such heterogeneous experiences into reliable skill updates. To address these issues, we present SkillClaw, a framework for collective skill evolution in multi-user agent ecosystems, which treats cross-user and over-time interactions as the primary signal for improving skills. SkillClaw continuously aggregates trajectories generated during use and processes them with an autonomous evolver, which identifies recurring behavioral patterns and translates them into updates to the skill set by refining existing skills or extending them with new capabilities. The resulting skills are maintained in a shared repository and synchronized across users, allowing improvements discovered in one context to propagate system-wide while requiring no additional effort from users. By integrating multi-user experience into ongoing skill updates, SkillClaw enables cross-user knowledge transfer and cumulative capability improvement, and experiments on WildClawBench show that limited interaction and feedback, it significantly improves the performance of Qwen3-Max in real-world agent scenarios.
Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.
Zhuoyun Yu, Xin Xie, Wuguannan Yao +4
♠Zhejiang University · ♣Ant Digital Technologies, Ant Group · ♢Zhejiang University - Ant Group Joint Laboratory of Knowledge Graph +1
The rapid advancement of Large Language Models (LLMs) has driven their deployment toward general-purpose harness environments, where LLM-based agents operate in sandbox workspaces and interact directly with file systems and shells. While numerous post-training pipelines exist for tool-use agents, their counterparts for harness-based agents remain scarce and are predominantly one-shot and teacher-dependent: a stronger external model synthesizes tasks and collects trajectories, the target agent is trained once, and evaluation results are consumed by human researchers rather than fed back into the loop. Such pipelines are inherently bounded by the capability of the external model and the scope of human prior, leaving no room for self-improvement. To address these limitations, we propose ClawLoop, a framework that enables self-improving without any external teacher model and with minimal prior dependencies. ClawLoop composes three phases into a self-sustaining cycle: Task Synthesis, which generates a diverse pool of training tasks via two complementary construction routines, applies diagnosis-conditioned difficulty growth followed by both agent-free and agent-based task validation; Training, which updates the agent via SFT and/or RL; and Diagnosis and Feedback, which analyzes the agent's rollout behaviors to produce a capability analysis report that will be used as feedback for task synthesis in the next iteration. Using Qwen3.6-27B as the loop model, ClawLoop yields consistent improvements on SkillsBench, Claw-Eval-Live, and ClawGym-Bench. The iteratively refined tasks are further transferred to smaller-scale Qwen3.5-4B and 9B models with gains across all benchmarks, validating data quality independently of the loop model.