agentsclimarketplace

Tokenlab cost routing

Skill hedging8563/tokenlab-skills/skills/tokenlab-cost-routing

Agent skills for TokenLab API integration, model picking, native endpoints, cost routing, and OpenAI-compatible migration

Install
npx -y skills add hedging8563/tokenlab-skills --skill tokenlab-cost-routing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Choose TokenLab models and fallback chains using public pricing, task fit, latency expectations, and native endpoint needs before writing production routing code.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.8 KB, as published. Nobody here has run it

TokenLab Cost Routing

Use this skill when a user asks how to reduce TokenLab cost, compare model prices, pick fallbacks, or route requests by quality, latency, and budget.

What this skill should deliver

  • A compact routing recommendation with exact public TokenLab model IDs.
  • A cost-aware fallback chain for the user's workload.
  • A catalog/pricing lookup path that can be rerun.
  • A note on which endpoint family each model should use.
  • Guardrails for when not to switch models because doing so would change output, safety, or request semantics.

Preferred approach

  1. Identify the workload and constraints:
    • chat, coding, agent loop, image, video, audio, embedding, rerank, translation, or multimodal
    • quality floor
    • latency target
    • budget or cost ceiling
    • native endpoint requirement
  2. Read live public catalog signals before recommending:
    • GET https://api.tokenlab.sh/v1/models
    • GET https://api.tokenlab.sh/v1/models?recommended_for=<scene>
    • GET https://api.tokenlab.sh/v1/models/:model
    • GET https://api.tokenlab.sh/v1/models/:model/pricing
  3. Build a chain with roles:
    • primary quality model
    • balanced default
    • fast fallback
    • budget fallback
  4. If the user asks for exact cost, compute from live pricing and their estimated token/media volume. State units and assumptions.
  5. For non-chat requests, inspect model details before changing parameters or endpoint family.

Output format

  • One sentence stating workload and assumptions.
  • A table with Route role, Model ID, Endpoint, Why, and When to fall back.
  • One catalog command and one pricing command.
  • A short implementation note for retries, rate limits, and user approval when quality would drop.

Avoid

  • Do not invent prices, discounts, or model counts.
  • Do not choose a cheaper model if that would silently remove required native behavior, tools, media support, safety constraints, or structured output guarantees.
  • Do not expose TokenLab internal channel, physical provider, or routing details.
  • Do not turn a user-provided model into a different model without saying why.
  • Do not hardcode a fallback list without saying when it was checked or how to refresh it.

Edge Cases

  • If catalog or pricing endpoints are unavailable, say that routing cannot be price-verified and provide only an example pattern.
  • If the user asks for "cheapest", include capability and reliability tradeoffs.
  • If billing risk is high, require explicit user approval before adding automatic fallback to paid media/video generation.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.