ClawRouter

Smart LLM router — save 67% on inference costs. Routes every request to the cheapest capable model across 41 models from OpenAI, Anthropic, Google, DeepSeek,...

Security Scan

The skill's instructions are mostly coherent with a model-routing utility, but it omits key details about required credentials/config and relies on an external plugin install (not included), which raises transparency and privilege concerns.

Purpose & Capability

Name/description match the instructions: it routes requests to many models and tells the agent to set models to blockrun/auto. However the description mentions routing 'all... through one wallet' while the skill declares no required env vars and only a single config path (models.providers.blockrun) — this is plausible if BlockRun is an aggregator, but the skill does not document what credentials or wallet config are required.

Instruction Scope

SKILL.md only instructs installing the external OpenClaw plugin and switching the agent's model to blockrun/auto or a pinned model. It does not ask the agent to read unrelated files, secrets, or system paths beyond the declared models.providers.blockrun config path.

Install Mechanism

This packaged skill is instruction-only and contains no code, but the provided workflow requires running 'openclaw plugins install @blockrun/clawrouter' which will download and install external code not present in this skill. The skill gives no install spec here, so installing will fetch code from outside the package (the homepage repo is given). That external install step could place arbitrary code on disk and should be audited before running.

Credentials

The skill declares no required environment variables but does require the config path models.providers.blockrun. The description's 'one wallet' claim implies sensitive credentials/payment details will be used. The skill does not enumerate what secrets (wallet keys, provider API keys) are needed or how they are stored, which is a transparency gap and a proportionality concern.

Persistence & Privilege

always is false and there are no indications the skill requests persistent system-wide privileges or modifies other skills' configs. Autonomous invocation is allowed (default) but not combined here with other high-risk flags.

What to consider before installing

This skill appears to be a router that relies on an external OpenClaw plugin (@blockrun/clawrouter). Before installing: (1) Inspect the external plugin's source (GitHub link) and release artifacts to see what code will be installed. (2) Check what configuration the 'models.providers.blockrun' entry contains — does it require a wallet key, API keys for many providers, or other secrets? If so, understand where those secrets are stored and who can access them. (3) Prefer installing in an isolated/test agent first to observe behavior and network activity. (4) Apply least-privilege: avoid supplying unrelated provider API keys or broad filesystem/agent permissions. (5) If you need stronger assurance, ask the maintainer for a clear list of required credentials and an audit or reproducible build of the plugin. These steps will reduce the risk of unintentionally exposing credentials or installing unreviewed code.

Smart LLM router that saves 67% on inference costs by routing each request to the cheapest model that can handle it. 41 models across 5 providers, all through one wallet.

Install

openclaw plugins install @blockrun/clawrouter

Setup

# Enable smart routing (auto-picks cheapest model per request)
openclaw models set blockrun/auto

# Or pin a specific model
openclaw models set openai/gpt-4o

How Routing Works

ClawRouter classifies each request into one of four tiers:

• SIMPLE (40% of traffic) — factual lookups, greetings, translations → Gemini Flash ($0.60/M, 99% savings)
• MEDIUM (30%) — summaries, explanations, data extraction → DeepSeek Chat ($0.42/M, 99% savings)
• COMPLEX (20%) — code generation, multi-step analysis → Claude Opus ($75/M, best quality)
• REASONING (10%) — proofs, formal logic, multi-step math → o3 ($8/M, 89% savings)

Rules handle 80% of requests in <1ms. Only ambiguous queries hit the LLM classifier ($0.00003 per classification).

Available Models

41 models including: gpt-5.2, gpt-4o, gpt-4o-mini, o3, o1, claude-opus-4.6, claude-sonnet-4.6, claude-haiku-4.5, gemini-3.1-pro, gemini-2.5-pro, gemini-2.5-flash, gemini-2.5-flash-lite, deepseek-chat, deepseek-reasoner, grok-3, grok-3-mini.

Example Output

[ClawRouter] google/gemini-2.5-flash (SIMPLE, rules, confidence=0.92)
             Cost: $0.0025 | Baseline: $0.308 | Saved: 99.2%