Skip to content
Get packs

Guide · Agent skill

How to use the Benchling Integration agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The benchling-integration skill is a set of instructions and reference files that teaches an AI agent how Benchling, a cloud platform many labs use for sequences, samples, inventory and electronic lab notebook entries, expects to be queried and updated. It is aimed at bioinformatics and lab software developers, and at lab data teams who export or audit data from a Benchling tenant. After this guide you will know how to install the skill, what it needs, how to ask for help with natural requests, and which safety checks to keep in place.

Benchling IntegrationK-Dense-AI/scientific-agent-skills · MIT

Quick start

  1. 1Download and unzip. Download the skill ZIP from this page and unzip it into ~/.claude/skills/ or your project's .claude/skills/ folder.
  2. 2Install the SDK. Run uv pip install "benchling-sdk==1.25.0" in the Python environment you will use.
  3. 3Set credentials. Set BENCHLING_TENANT_URL and either BENCHLING_API_KEY or BENCHLING_CLIENT_ID with BENCHLING_CLIENT_SECRET.
  4. 4Run a read-only check. Ask the agent for a small filtered list query and confirm it returns a page without errors.
  5. 5Make a natural request. Describe the task, such as importing sequences or auditing containers, and name the tenant and folder or schema.
  6. 6Review before writing. Check the schema IDs and payload the agent proposes, and run writes against a test tenant first.
  7. 7Read back results. Ask the agent to fetch the created or updated records and compare them with what you intended.

What is an agent skill and what does this one do?

An agent skill is a folder of instructions and reference files that an AI agent loads when a task matches its description. This one is named benchling-integration and its SKILL.md covers Benchling registry entities, sequence imports, inventory, ELN entries, workflow tasks, apps, event-driven integrations and warehouse analytics.

Once installed, the agent writes code with the benchling-sdk 1.25.0 library and the stable v2 API. It follows a safer workflow: identify the tenant and permissions, read the schema and resolve real folder and dropdown IDs, read a small filtered page before writing, serialize a representative payload, then read the result back. It keeps external IDs and returned Benchling IDs so a retry does not create duplicates.

Install the benchling-integration skill

Download the ZIP for this skill from this page and unzip it. For a personal setup, place the folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.

Alternatively, upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

  • Personal: ~/.claude/skills/benchling-integration
  • Project: .claude/skills/benchling-integration
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Set up Python, the SDK and your credentials

The examples need Python 3.9 or newer and benchling-sdk 1.25.0, which the skill installs with uv or pip. Set BENCHLING_TENANT_URL, which is required. Then set either BENCHLING_API_KEY, or the OAuth pair BENCHLING_CLIENT_ID and BENCHLING_CLIENT_SECRET. Multi-environment setups can also use BENCHLING_PROD_ and BENCHLING_STAGING_ tenant URLs and API keys.

The skill's first runnable example is a read-only list query that checks connectivity. An empty page is a valid result, but it does not prove you can see every project.

uv pip install "benchling-sdk==1.25.0"

Trigger the skill with natural requests

You do not need special syntax. Describe the Benchling task in plain language and mention Benchling, and the agent should load the skill when the request matches its description. Good requests name the object and the outcome, for example importing a FASTA file into a folder, or listing containers under a box.

Tell the agent your tenant URL variable, which identity you use, and whether this is a test tenant. The skill asks you to confirm schemas and permissions before writes.

  • Write me a script that lists DNA sequences in a schema using benchling-sdk 1.25.0.
  • Import the sequences in sequences.fasta into folder lib_example and save a mapping of names to Benchling IDs.
  • Audit all containers under storage box box_example and print their barcodes.

Worked examples the skill covers

The skill includes three illustrative patterns. Each one is a starting point that must be checked against your tenant before use.

FASTA import creates unregistered linear DNA sequences. It needs Biopython, installed separately, and you should choose topology and resolve name collisions first.

Inventory audit lists containers under a location. The ancestor filter includes descendants, so immediate children need a separate check.

CSV export writes the id, name, bases and length of sequences in one schema to a file.

uv pip install biopython

Safety: scripts, network access and test tenants

This skill runs scripts and uses the network. The examples call your Benchling tenant over HTTPS with your credentials, so run them with credentials that have only the permissions you intend to use. Review generated code before running it.

The skill states that its tenant-dependent examples were illustrative and were not run against a live tenant. The examples create and update real records, so use a test tenant and get a review before any production write. The skill does not check your tenant's schemas or the scientific correctness of your data.

  • Keep secrets in environment variables, not in code.
  • Use a test or staging tenant for first runs.
  • Read back results and check async status, since a finished poll can still show FAILED.

What people use it for

  • Importing a FASTA file of sequences into a Benchling folder while saving the mapping to Benchling IDs.
  • Auditing which containers and barcodes sit under a storage box.
  • Exporting sequences from one schema to a CSV file for analysis.
  • Maintaining an existing benchling-sdk v2 integration and checking that method names match SDK 1.25.0.

Frequently asked questions

Is the Benchling Integration agent skill free?

The skill is published under the MIT licence, as listed on this page. Benchling itself is a separate commercial platform, and you need your own tenant with API access to use the examples.

What do I need to use the benchling-integration skill?

You need Python 3.9 or newer, benchling-sdk 1.25.0, network access, a Benchling tenant with API access, and either an API key or OAuth app credentials. BENCHLING_TENANT_URL is required. Biopython is only needed for the FASTA import example.

How do I install the benchling-integration skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

Can the skill run against my live Benchling tenant?

It can write code that does, but the skill's examples were not run against a live tenant. They create and update real records, so test on a non-production tenant and review the code first.

Why does my Benchling connection test return an empty list?

An empty page can be a valid result that means no matching items are visible to your credentials. Check that your app or key has access to the required projects, teams or organizations.

How does this compare with using the Benchling SDK directly?

The skill gives an agent the SDK conventions, safer write steps and examples so you do not have to look them up. You still need to verify schemas and results yourself, and the skill does not cover non-Benchling lab systems.

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

This skill suits bioinformatics and lab software developers, and lab data teams, who need to script Benchling sequence, inventory or export tasks with benchling-sdk 1.25.0 and the v2 API. It is not a good fit if you have no Benchling tenant and credentials, if you want to write to production without a test tenant and review, or if your lab uses a non-Benchling system.

Download the skill