agentsclimarketplace
Sign in

Tokenlab model picker

Skill hedging8563/tokenlab-skills/skills/tokenlab-model-picker

Pick TokenLab models for chat, coding, image, video, audio, embeddings, reranking, and translation by reading public model catalog signals before recommending concrete model IDs.From its SKILL.md

Install
npx -y skills add hedging8563/tokenlab-skills --skill tokenlab-model-picker

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

2 things 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.
  • fetches URLsInstructs the agent to fetch 4 URLs, including https://api.tokenlab.sh/v1/models and 3 more.

What its file declares

Copied from the file, not written here

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

2.9 KB, 596 tokens by cl100k_base, as published. Nobody here has run it

TokenLab Model Picker

Use this skill when a user asks which TokenLab model to use, how to compare model options, or how to route a workload across model families.

What this skill should deliver

  • A short model shortlist with exact TokenLab model IDs.
  • The workload assumptions used to pick the models.
  • A public catalog lookup path that the user or agent can rerun.
  • A fallback model when the first choice is unavailable or too expensive.
  • A caveat when a recommendation depends on volatile pricing, availability, or benchmark data.

Preferred approach

  1. Identify the workload: chat, coding, agent loop, image, video, audio, embedding, rerank, translation, or multimodal.
  2. Use the public model catalog before recommending hardcoded IDs:
    • General catalog: GET https://api.tokenlab.sh/v1/models
    • Task shortlist: GET https://api.tokenlab.sh/v1/models?recommended_for=<scene>
    • Model contract: GET https://api.tokenlab.sh/v1/models/:model
    • Pricing detail: GET https://api.tokenlab.sh/v1/models/:model/pricing
  3. Prefer exact public model IDs over family names.
  4. Separate recommendation dimensions:
    • quality or frontier capability
    • cost sensitivity
    • latency or fast iteration
    • native endpoint needs
    • multimodal input or output
  5. Return a compact table, then one runnable API example if useful.

Default shortlist patterns

  • Coding and agent work: choose a strong reasoning/coding model, a cheaper fallback, and a fast iteration model.
  • General chat: choose one balanced model and one lower-cost fallback.
  • Image or video: use recommended_for=image or recommended_for=video instead of guessing request shapes.
  • Embeddings, rerank, translation, TTS, STT, music, or 3D: use the task-specific shortlist and inspect the model contract before showing parameters.

Output format

  • One sentence naming the workload assumptions.
  • A table with Use, Model ID, Why, and Fallback.
  • One catalog command the user can rerun.
  • One warning line if availability, pricing, or provider-native behavior must be verified.

Avoid

  • Do not claim a single universal best model.
  • Do not recommend provider-prefixed or physical route names as public model IDs.
  • Do not invent prices or model counts.
  • Do not silently translate a native-only need into a generic chat completion.
  • Do not recommend a model that is absent from the current public catalog.

Edge Cases

  • If the user asks for the cheapest option, still include capability limits.
  • If the user asks for a benchmark winner, require a cited benchmark and observed date.
  • If the catalog is unavailable, say so and fall back to the last known examples only as examples, not truth.

What ships with it: 2 files

558 B alongside SKILL.md

agents/

Gives 0 of the 12 instructions most video audio skills give in 596 tokens

Counted across 619 of the 725 authors here whose files we hold, read 2026-09-06

  • Read product marketing context firstin 13 of 619, across 7 files
  • Define the core visual thesis in one sentencein 11 of 619, across 3 files
  • Break the concept into 3 to 6 scenesin 11 of 619, across 3 files
  • Render the smallest working version firstin 11 of 619, across 3 files
  • Start with a low-quality smoke test renderin 11 of 619, across 3 files
  • Add captions for accessibility and engagementin 11 of 619, across 5 files
  • Write the scene outline before writing codein 11 of 619, across 3 files
  • Specify subject, action, camera, style, and moodin 11 of 619, across 5 files
  • Decide what each scene provesin 10 of 619, across 2 files
  • Export one clean thumbnail framein 10 of 619, across 2 files
  • Pick the right tool for the jobin 10 of 619, across 4 files
  • Run the test suite before proposing a fixin 8 of 619, across 7 files

Said here and by no other author read

  • Identify the workload
  • Use the public model catalog
  • Prefer exact public model IDs
  • Separate recommendation dimensions
  • Return a compact table

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. 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.