Your agent's burning tokens on retry loops, bloated prompts, and expensive models doing lightweight work. ClawClip shows you where — so you can cut the waste and keep what actually makes it stronger.
OpenClaw's built-in Usage Dashboard shows you what you spent. ClawClip shows you why you spent it, whether your changes helped, and how to spend less.
ClawClip fills the gaps OpenClaw doesn't cover: step-by-step session replay, before/after version comparison, cross-framework analysis (Hermes, LangGraph, AutoGen), and actionable optimization recommendations.
Runs locally · Reads your existing logs · No upload · No code changes
Live Demo · Quick Start · Preview · FAQ · What it does · Roadmap · 中文 · 日本語 · 한국어 · Español · Français · Deutsch
You're running OpenClaw or ZeroClaw. Token costs keep climbing. You swap models, tweak prompts, and hope it helped — but you're never sure.
ClawClip ends the guessing. It reads your session logs, replays what actually happened, scores the result, and shows you exactly where the money went. Cut the waste. Pick models that earn their cost. Keep only the changes that make your agent genuinely stronger.
Retry loops you didn't notice. Prompts that grew too long. Expensive models doing simple work. ClawClip reads your logs, flags the patterns, and tells you which fix saves the most.
Swap a model, tweak a prompt, run it again. ClawClip shows you the before/after: score, tokens, cost. If it got better but more expensive, you'll see that too.
Found a waste pattern? ClawClip suggests free-tier alternatives and generates the config for you. No digging through docs.
The Agent Run Diagnostics is a heuristic diagnostic — not a standardized benchmark. It reads signals from your actual sessions (response quality, tool use, safety patterns, cost structure) to help you compare iterations faster. Use it for relative improvement tracking, not absolute rankings.
| What you actually want to know | What ClawClip shows you |
|---|---|
| Where's my token budget going? | Cost Report breaks spend by model, task, and session — with waste signals and savings suggestions |
| Is my agent actually getting better? | Agent Run Diagnostics reads behavioral signals from your sessions (tool use patterns, retry loops, cost structure) to help you spot regressions faster. Use it for relative tracking, not absolute rankings. |
| What exactly happened in that run? | Run Insights replays every step so you can find the problem without reading raw logs |
| Feature | What it's for |
|---|---|
| Token waste detection | Flags retry loops, context bloat, prompt inefficiency, and model mismatches |
| Model value matrix | Shows which models deliver the best results per dollar across your actual tasks |
| Before/after proof | Compares the last two benchmark runs with a plain verdict |
| Savings suggestions | Prioritizes the changes most likely to cut cost without hurting quality |
| Prompt Efficiency | Checks whether longer prompts and more tokens are actually buying better output |
| Version Compare | Side-by-side comparison of different runs, models, or configs |
| Template Library + Knowledge Base | Reuse what worked, search your history, stop repeating the same experiments |
- Session discovery, parsing, and all analysis run locally
- ClawClip does not upload your agent run data
- Pricing refresh is optional — it only updates cost reference numbers, never sends session content
git clone https://github.com/Ylsssq926/clawclip.git
cd clawclip && npm install
npm startOpen http://localhost:8080. The built-in demo sessions load immediately — no setup needed. When you're ready, point ClawClip at your own OpenClaw or ZeroClaw logs.
Primary support: OpenClaw and ZeroClaw official session formats.
Also works with: any local JSONL-based agent workflow — coverage expands as the parser grows.
# Point at a custom log directory
CLAWCLIP_LOBSTER_DIRS=/path/to/your/sessions npm start| Source | Notes |
|---|---|
~/.openclaw/ |
Auto-discovered at startup |
OPENCLAW_STATE_DIR |
Override the default OpenClaw state path |
CLAWCLIP_LOBSTER_DIRS |
Add extra folders (comma or semicolon separated) |
| Built-in demo sessions | Available immediately, no real data needed |
| ZeroClaw exports / other JSONL | Supported progressively |
| OpenTelemetry (OTLP/HTTP) | LangGraph / AutoGen via OTEL exporter (see below) |
ClawClip can receive traces from LangGraph and AutoGen via OpenTelemetry.
pip install langsmith[otel]
export LANGSMITH_OTEL_ENABLED=true
export OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=http://localhost:8080/api/otel/v1/traces
export OTEL_EXPORTER_OTLP_TRACES_PROTOCOL=http/jsonfrom opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import SimpleSpanProcessor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
provider = TracerProvider()
provider.add_span_processor(
SimpleSpanProcessor(OTLPSpanExporter(endpoint="http://localhost:8080/api/otel/v1/traces"))
)- Shareable reports: export a full Replay + Scorecard + Cost summary as a static snapshot
- AI-assisted diagnosis: LLM-powered second opinion on top of the existing heuristics
- Broader agent framework support: more local JSONL runtimes beyond OpenClaw / ZeroClaw
- Real-time monitoring: live session ingestion as runs happen
- QQ Group:
892555092 - GitHub Discussions: Ask a question or share what you built
- Issues: Report a bug or request a feature
Built with 🍤 by Luelan (掠蓝)
If ClawClip helped you cut your token bill or find a better model config, a ⭐ goes a long way.