adaptyv

Trigger phrases

Phrases that activate this skill when typed to Claude Code:

• Adaptyv
• Foundry API
• protein binding assay
• submit protein sequences for characterization

What it does

adaptyv is a Claude Code skill from K-Dense AI’s scientific-agent-skills repo. It turns Claude into a hands-on assistant for Adaptyv Bio’s cloud foundry platform — helping you design protein experiments, submit sequences via the REST API or Python SDK, and retrieve assay results programmatically.

The output of a session is working Python code that authenticates against the Foundry API, submits amino acid sequences for experimental characterization (binding, thermostability, expression, fluorescence), and polls for results — all without hardcoding credentials or leaving the terminal.

Adaptyv’s automated lab runs assays in approximately 21 days. The skill handles the protocol-side glue so you can focus on which proteins to test rather than how to talk to the platform.

When to use it

Reach for it when:

• You have a set of designed or engineered protein sequences and want to submit them for BLI, SPR, or thermostability assays at Adaptyv Bio
• You need to build a programmatic pipeline that polls the Foundry API for results and feeds them back into a design loop
• You are setting up environment variables and SDK configuration for a new Adaptyv project and want boilerplate that avoids credential leaks

When not to reach for it:

• You need to analyze or visualize already-returned structural data — reach for a structural biology tool instead
• Your sequences are not yet designed and you need a generative model — pair with the esm skill first

Install

Copy the SKILL.md from scientific-skills/adaptyv into .claude/skills/adaptyv/.

The skill activates on trigger phrases including “Adaptyv”, “Foundry API”, “protein binding assay”, and “submit protein sequences for characterization”.

What a session looks like

A typical session has three phases:

1. Environment setup. Claude reads your project root for a .env file and configures ADAPTYV_API_KEY and ADAPTYV_API_URL using python-dotenv. No token touches source control.
2. Experiment submission. Claude generates SDK calls using the decorator pattern from adaptyv-sdk, targeting the correct assay type and submitting your sequence list with appropriate metadata fields.
3. Result retrieval. Claude writes a polling loop or one-shot fetch that returns assay results as structured data (binding scores, Tm values) ready for downstream analysis or re-ranking.

Receipts

Honest reporting on what adaptyv handles well and where it falls short:

Where it works well:

• Generating boilerplate API code for first-time Foundry integrations — saves reading through scattered SDK docs
• Multi-sequence batch submission scripts where the same assay parameters apply to a panel of variants
• Setting up CI-style result polling for design-build-test loops

Where it backfires:

• The skill cannot simulate assay outcomes — results still take the full lab turnaround cycle
• If the Foundry API schema changes between SDK versions, generated code may reference deprecated fields; always pin the SDK version

Pattern that works: run the submission code in a dry-run mode first (log the request payload without posting) to verify sequence formatting and metadata before spending assay budget.

Source and attribution

Originally authored by K-Dense, Inc.. The canonical SKILL.md lives in the adaptyv folder of the scientific-agent-skills repository.

License: MIT. Install, adapt, and redistribute with attribution preserved.

This page documents the skill from a practitioner’s perspective. For the formal spec and updates, defer to the source repo.