Skip to main content
The Tamarind CLI (tamarind) lets you discover tools, submit and monitor protein, nucleic-acid, and small-molecule jobs, and download results — from your terminal, a shell script, CI, or an AI coding agent such as Claude Code or Codex. It is a thin client over the same REST API as the Tamarind MCP server, so the two stay in lockstep: every tool and parameter available in the API is available from the CLI. Source, installer, and issues live on GitHub. Use the CLI when you want to drive Tamarind from a terminal or a script. Use the MCP server when you want a hosted connection for claude.ai, ChatGPT, or another MCP client.

​
Requirements

  1. A Tamarind account — Sign up here
  2. An API key — view yours at app.tamarind.bio/api-docs/api-key
  3. Python 3.10 or newer

​
Installation

Install once with uv or pipx, then run tamarind from anywhere: 
uv tool install tamarind-cli
# or:
pipx install tamarind-cli
You can also use the bootstrap installer (it installs uv first if needed): 
curl -fsSL https://app.tamarind.bio/cli/install.sh | sh
To track the latest from source: 
uv tool install "git+https://github.com/Tamarind-Bio/tamarind-cli"
Verify the install: 
tamarind --version

​
Authentication

Get your API key from app.tamarind.bio/api-docs/api-key, then authenticate in one of two ways. Environment variable — best for scripts, CI, and AI agents: 
export TAMARIND_API_KEY="YOUR_API_KEY_HERE"
Stored config — saves the key to ~/.tamarind/config.json: 
tamarind auth login
tamarind auth status

​
Quickstart

The end-to-end flow mirrors the platform: discover a tool, inspect its schema, validate your inputs, submit, and download results. 
# 1. Find a tool
tamarind tools --function structure-prediction --modality protein
tamarind tools --search boltz

# 2. Inspect its parameters and grab a runnable example
tamarind schema boltz
tamarind schema boltz --example > job.yaml

# 3. Validate, then submit and wait for results
tamarind validate boltz --input job.yaml
tamarind submit   boltz --input job.yaml --name my-run --wait --download ./out

# 4. Monitor anytime
tamarind jobs
tamarind status my-run
tamarind results my-run --download ./out

​
Command reference

​
Discover

CommandDescription
tamarind toolsList available tools (filter with --modality, --function, --search)
tamarind modalitiesList modalities (protein, nucleic acid, small molecule, …)
tamarind functionsList tool functions (structure prediction, binder design, …)
tamarind schema <tool>Show a tool’s parameters; --example prints a runnable job file

​
Submit

CommandDescription
tamarind validate <tool> --input job.yamlValidate inputs without submitting
tamarind submit <tool> --input job.yamlSubmit a job (--name, --wait, --download, --idempotency-key)
tamarind batch <tool> --input batch.yamlSubmit many jobs as one batch

​
Monitor & results

CommandDescription
tamarind jobsList jobs (--status, --batch, --limit)
tamarind status <jobName>Show one job’s status
tamarind wait <jobName>Block until a job reaches a terminal state
tamarind logs <jobName>Fetch a job’s logs
tamarind results <jobName> --download ./outDownload a job’s result bundle

​
Files & lifecycle

CommandDescription
tamarind files listList workspace files
tamarind files upload <path>Upload an input file
tamarind files delete <path>Delete a file
tamarind cancel <jobName>Stop a running or queued job (--batch for a whole batch)
tamarind delete <jobName>Permanently delete a job and its data

​
Output & scripting

Every command is built for automation:
  • --json prints machine-readable output. It is on by default when output is piped (not a terminal), so scripts and agents get JSON without any extra flags.
  • Stable exit codes let scripts branch on failures: 0 success, 3 auth error, 4 not found, 5 validation error, 6 rate limited.
  • --profile <name> switches between saved accounts or environments in ~/.tamarind/config.json.
  • --api-key, --api-base, and --catalog-base (or the matching TAMARIND_* environment variables) override configuration per invocation.

​
Use with AI agents

The CLI is built for coding agents. Because every command takes plain flags, emits JSON, and returns stable exit codes, an agent can drive the whole workflow — discover a tool, write the job file, submit, and fetch results — by running tamarind directly.
  • Terminal agents (Claude Code, Codex, CI): install the CLI, set TAMARIND_API_KEY, and let the agent call tamarind commands. Point it at this page so it knows the surface.
  • Hosted / chat agents (claude.ai, ChatGPT): connect the Tamarind MCP server, which exposes the same operations over the Model Context Protocol.
A packaged plugin with task-specific skills (antibody design, binder design, batch screening) is in the works — contact info@tamarind.bio if you’d like early access.

​
Troubleshooting

​
No API key configured

Set TAMARIND_API_KEY or run tamarind auth login. Confirm with tamarind auth status.

​
Authentication failing

Verify your key is current on the API Key page. Pass --api-key directly to rule out a stale stored key.

​
command not found: tamarind

Ensure your tool installer’s bin directory is on PATH (for uv, run uv tool update-shell; for pipx, run pipx ensurepath), then restart your shell. Need help? Contact us at info@tamarind.bio.