Conciseness

Sections like Data Flow, Credentials & Local Key Storage, Install, and Setup are tight and high-signal, but the ~60-model inline catalog and time-sensitive minutiae ('~20,256 tokens/sec at the 720p+audio default — 2× the prior 480p rate', 'since 2026-06-01', 'Router Core V3.4') are padding that churns and belongs in a reference file. Mostly efficient but should be tightened — not a 2 because most sections earn their tokens.

Actionability

Concrete, executable commands throughout: 'openclaw plugins install @blockrun/clawrouter', 'openclaw models set blockrun/auto', the curl commands for the live model catalog, '/cr-call +1<E.164> "<task>"', 'clawrouter phone numbers list/buy/renew/release', and the 'npm pack @blockrun/clawrouter@<version>' verification, plus concrete pricing formulas and tool parameters. Minor gap: no example of an actual LLM call made through the localhost proxy, so it stops short of fully covering the most common case.

Workflow Clarity

Install → setup → routing is unambiguous, and the real-money operations carry explicit validation checkpoints: Polymarket requires 'confirm:true' to place/sign anything (omit → dry-run preview) with 'POLYMARKET_MAX_BET_USD' caps, voice calls are flagged as real-world side effects with 'only call blockrun_voice_call when the user has explicitly asked', and the voice polling loop (poll every 10–30s) is stated. Not a 5: some flows (e.g. first wallet funding, x402 setup) defer to other skills without an inline checkpoint summary.

Progressive Disclosure

Section headers are clear and references are signaled (links to the phone, predexon, and polymarket-trading skills and to GitHub files), but no bundle files exist — the 90-model catalog and the full 8-mode billing table are inlined content that clearly belongs in separate reference files. The 'docs/routing-profiles.md' link uses a '../../docs/' path that points outside the skill bundle, and the referenced sibling skills are not part of this bundle.

Specificity

The description lists many concrete capabilities ('forwards each request to the blockrun.ai gateway, which routes to the cheapest capable model', 'realtime market data (global stocks, crypto, FX, commodities)', 'SIM-swap fraud detection', 'AI-powered outbound voice calls (Twilio + Bland.ai)'), but it is padded and includes the over-claim 'save 84% on inference costs'. Not a 5: verbosity is penalized and minor coverage gaps exist (e.g. 'Twitter/X intelligence' is claimed but not substantiated in the skill body).

Completeness

The 'what' is clearly answered ('a local proxy that forwards each request to the blockrun.ai gateway, which routes to the cheapest capable model'), but there is no 'Use when...' clause or equivalent explicit trigger guidance inside the description — the boundary note 'Not a local-inference tool — prompts are sent to the blockrun.ai gateway' only weakly implies when to use it, capping completeness at 3 per the judging guidelines.

Trigger Term Quality

Natural domain terms are present ('LLM router', 'cheapest capable model', 'free open-weight models', 'prediction-market', 'SIM-swap fraud detection', 'outbound voice calls'), giving good keyword coverage. Not a 5: common variations users would actually say (e.g. 'OpenRouter alternative', 'cheap LLM API') appear only in the separate triggers field, not in the description itself.

Distinctiveness Conflict Risk

The blockrun.ai gateway niche is distinct, but the description sprawls across stocks, crypto, FX, commodities, Twitter/X intelligence, prediction markets, phone-number verification, SIM-swap detection, and voice calls — each a domain where a dedicated skill would create real trigger overlap. It fits 'somewhat specific but could still overlap with similar skills' better than the minor-overlap anchor.