Content
50%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is dense with genuinely useful, product-specific guidance and concrete commands, tables, and safety gating for the real-money operations. Its weaknesses are a marketing-heavy, token-inefficient tone, inlined catalogs that belong in reference files, missing verification steps in the setup workflow, and references to skill files that are not present in the bundle.
Suggestions
Trim marketing prose ("killer feature", "pattern for new APIs going forward", the savings-mix justification) and move the ~40-model catalog and per-tool price tables into a reference file, keeping only tier defaults and the free-tier caveat inline.
Add a verification step to the setup workflow (e.g. a curl or SDK test call against the local proxy with expected Example Output) so the install → setup → use sequence has a checkpoint.
Show the proxy's chat-completions base URL and one minimal end-to-end request example, mirroring the Surf section's `http://127.0.0.1:8402/v1/surf/...` pattern.
Fix the `../../docs/routing-profiles.md` reference to a path inside the skill bundle, or inline the tier chains it was meant to provide.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Marketing prose and padded sections recur ("The killer feature is…", "This is the pattern for new BlockRun-marketplace APIs going forward", the savings-mix disclaimer paragraph) and the "Available Models" section is a wall of ~40 model names with inline prices — time-sensitive data with no old-patterns/deprecated section. Content is product-specific rather than re-teaching known concepts, so it is not a 1, but it is noticeably verbose. | 2 / 5 |
Actionability | Copy-pasteable commands appear throughout — `openclaw plugins install @blockrun/clawrouter`, `openclaw models set blockrun/auto`, `/cr-call +1<E.164> "<task>"`, `clawrouter phone numbers list`, poll cadence "every 10–30s" — plus tool tables with exact parameters. Minor gap: no example of an actual LLM call through the proxy (the chat base URL is never shown; only the Surf namespace `http://127.0.0.1:8402/v1/surf/...` is given). | 4 / 5 |
Workflow Clarity | Install → setup → use is sequenced, and the Polymarket section has validation-like gating (`confirm:true` hard-required, dry-run default, $25 per-order cap). But core workflows lack verification checkpoints — nothing verifies the proxy is running or demonstrates a test request — and the guideline caps workflow clarity at 3 for destructive/real-money and batch operations without full validation loops. | 3 / 5 |
Progressive Disclosure | Sections are well-organized and details are deferred to clearly signaled sibling skills ("See the `predexon` skill for the full endpoint reference"), but none of those skills exist in this bundle, `../../docs/routing-profiles.md` points outside it, and heavy catalog content (the model list, full tool/price tables) is inlined in SKILL.md rather than in reference files. | 3 / 5 |
Total | 12 / 20 Passed |