Guide · Agent skill
How to use the Tamarind Bio Molecular Tools agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The tamarind skill gives an AI agent the knowledge to run open-source molecular design and structural biology tools on the Tamarind Bio cloud platform, so you do not need a local GPU. It is for researchers, computational biology teams and developers who want to run tools such as AlphaFold, Boltz, RFdiffusion, ProteinMPNN and DiffDock from a conversation. This guide explains what the skill does, how to install it, how to trigger it, and what you need before it can run jobs.
Tamarind Bio Molecular ToolsK-Dense-AI/scientific-agent-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the tamarind skill ZIP from this page and unzip it. Keep SKILL.md and the references folder together.
- 2Place or upload the skill. Put the folder in ~/.claude/skills/ or in a project's .claude/skills/, or upload it under Settings, Capabilities, Skills in the Claude apps.
- 3Set your API key. Create an account, get an API key from your deployment's API settings, and store it in the TAMARIND_API_KEY environment variable.
- 4Ask a scientific question. Describe the task and your input, for example a protein sequence and the structure prediction you want, and mention Tamarind.
- 5Review and verify results. Let the agent poll the job and download results, then check the outputs scientifically before you rely on them.
What the tamarind skill makes the agent do
An agent skill is a folder of instructions and reference files that an AI agent loads when a task matches the skill. The tamarind skill covers Tamarind Bio, a platform that runs open-source molecular and structural biology tools on managed compute, accessed through its REST API or MCP server.
The skill covers structure prediction (AlphaFold, Boltz, Chai, ESMFold), protein, binder and de novo design (RFdiffusion, ProteinMPNN, BoltzGen), antibody and nanobody design and developability, protein-ligand docking (DiffDock, Autodock Vina), binding-affinity prediction, MSA generation and molecular dynamics. Once installed, the agent can look up available tools, check job settings, submit jobs with your API key, poll for completion and download results.
- Structure prediction and cofolding
- Binder and de novo protein design
- Inverse folding and sequence design
- Small-molecule docking and binding-affinity prediction
- MSA generation and molecular dynamics
Install the tamarind skill in Claude
Download the ZIP for the tamarind skill from this directory page and unzip it. For a personal install, place the unzipped skill folder inside ~/.claude/skills/. For a single project, place it inside that project's .claude/skills/ folder.
You can also upload the skill in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the skill there.
The skill folder contains SKILL.md and four reference files: references/api_reference.md, references/examples.md, references/tool_catalog.md and references/workflows.md. Keep these together so the agent can read them.
- Personal install: ~/.claude/skills/tamarind
- Project install: .claude/skills/tamarind inside your project
- Claude apps: Settings, then Capabilities, then Skills, then upload
Set up your Tamarind account and API key
The skill cannot run jobs without a Tamarind Bio account and an API key from app.tamarind.bio. Get the key from your deployment's API settings. If your organization uses a dedicated deployment, it has its own host and account data, so use that host instead of the shared one.
Store the key in the TAMARIND_API_KEY environment variable. The skill sends it as the x-api-key header. Do not put keys in files or logs. The skill needs Python 3.10 or later and the requests library in the environment where the code runs, and network access to app.tamarind.bio.
Free usage is a monthly allowance, not an unconditional promise. Check your account's current allowance and compute budget before you scale up.
- Python 3.10 or later
- requests library
- Tamarind Bio account and API key
- TAMARIND_API_KEY environment variable
- Network access to app.tamarind.bio
curl --fail-with-body 'https://app.tamarind.bio/tools.json?type=alphafold'Trigger the skill with natural requests
You do not need special syntax. The skill loads when your request mentions Tamarind or tamarind.bio, asks to run one of the listed open-source tools in the cloud, references app.tamarind.bio/api or the x-api-key header, or asks to submit batches of sequences for structural or biophysical characterization.
Name the scientific task and the input you have. The agent then matches the task to a tool, reads the tool's settings schema and validates the job before submitting it. The skill steers the agent through discover, read schema, validate, submit once, poll, then download and inspect.
- Predict the structure of this protein sequence on Tamarind using AlphaFold.
- Dock this ligand into a fixed receptor with Autodock Vina on Tamarind Bio.
- Run ProteinMPNN on this backbone structure through Tamarind and tell me the status of the job.
- Check which Tamarind tools can do antibody design and list their required settings.
Read results and check them scientifically
Results are downloaded as an archive for successful runs. For stopped or failed jobs, the skill requests the output.log file so you can see what went wrong. If a result is still being prepared, the agent retries retrieval without resubmitting the compute job.
The skill is explicit that it does not establish scientific accuracy for any model. Always verify the scientific outputs yourself. For design tasks, the skill recommends re-folding designs and comparing them to the intended backbone or interface.
- Complete: download the result archive
- Stopped or failed: request output.log
- Preparing: retry the result request later
Safety notes: scripts, network and data
This skill runs scripts and uses the network. It sends your API key to Tamarind and submits your inputs to a third-party cloud platform. Do not upload sensitive or unpublished sequences unless your data-confidentiality rules allow it.
For local sequence processing or molecular descriptor work, the skill itself says to use a local library instead.
- Scripts run: yes
- Network use: yes, to app.tamarind.bio
- Keep the API key out of files and logs
- Do not send the key when downloading signed result URLs
What people use it for
- Predicting protein or complex structures with AlphaFold, Boltz or ESMFold without local GPUs.
- Docking small molecules into a fixed receptor with Autodock Vina or DiffDock.
- Submitting batches of sequences for structural or biophysical characterization and tracking their status.
Frequently asked questions
How do I install the tamarind agent skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for a personal install, or into a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings, Capabilities, then Skills.
What do I need to use the tamarind skill?
You need Python 3.10 or later, the requests library, a Tamarind Bio account, an API key stored in TAMARIND_API_KEY, and network access to app.tamarind.bio. The skill cannot run jobs without credentials.
Is the tamarind skill free?
The skill is MIT licensed. Tamarind usage includes a monthly allowance, but the source does not promise unlimited free jobs, and billing and entitlements can change, so check your account's current allowance.
Why does the agent say my job is preparing or fails to find a tool?
A result request can return 202 with status preparing, so retry retrieval without resubmitting. If a tool is missing from the list, the skill notes that current custom deployments can be absent from the list, so check the known tool name and its schema first.
Can I use the tamarind skill without a GPU?
Yes. The jobs run on Tamarind Bio's managed compute, so you do not need a local GPU. You do need network access and an account.
Are there alternatives for local sequence work?
Yes. The skill says local sequence processing and molecular descriptor work should use a local library instead of the cloud platform.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the tamarind skill if you want an agent to run AlphaFold, Boltz, RFdiffusion, ProteinMPNN, DiffDock or Vina on Tamarind Bio without local GPUs, and you have an account and API key. Pick something else if you cannot upload your sequences to a third-party cloud, if you have no Tamarind account, or if you need validated clinical or safety conclusions, since the outputs still need expert review.
Download the skill