TL;DR
Every tool the platform can call, filtered by modality and function, each labelled by where it runs and how far it has been verified.
Use it / Skip it
Catalog is the single list behind every plan step. A tool here is a contract — inputs, outputs, and the executor it is bound to — not a page of hand-written glue. The badge on each card says where the work happens: Local, API, or Heuristic.
Use when
You want to know what is available for a modality, or to check what a plan step will actually call before you approve it.
Don't use for
Do not read imported as working. Imported means the schema is known; verified means we have run it and checked the result against the tool's own schema.
Inputs
- Search
- Free text over tool name and description.
- Modality filter
- Protein, antibody, peptide, small molecule and so on.
- Function filter
- Structure prediction, inverse folding, developability, de novo design.
Outputs
- Tool card
- Name, one-line description, execution badge, and its modality/function tags.
- Counts
- How many tools are verified against how many are imported.
Walkthrough
1. Filter by modality or function, or search by name
Filter by modality or function, or search by name.
2. Read the execution badge
Read the execution badge. Local costs nothing to run; API submits a metered job.
3. Check the verification level before relying on a tool inside a plan
Check the verification level before relying on a tool inside a plan.
Under the hood
- Tool definitions and their executor bindings are data, so adding a tool does not mean writing a new page.
- Verification level is recomputed from evidence on load, not stored as a hand-set flag.
Worked example
Pitfalls
- A Heuristic badge means a fast approximation for triage, not a substitute for the real tool.
- An imported tool has never been run here; treat its schema as documentation, not as evidence.