Sequencing Machines
An agent built on Vercel AI SDK to parse sequencing protocol and extract important details for downstream anlaysis, and a slack chatbot with chat SDK as well.
Core trick
Symbolic molecule schema with ordered top and bottom strand segments, exposed as deterministic skills via an MCP server.
Description
An agent built on Vercel AI SDK to parse sequencing protocol and extract important details for downstream anlaysis, and a slack chatbot with chat SDK as well.
README excerpt
# cdna Deterministic symbolic molecular skills for sequencing-library workflows, available as a Python library and an MCP server. ## Molecule schema in ten lines 1. `MoleculeState` has `id`, `label`, `strands`, `origin`, `evidence`, `skill_call_id`, `review_status`, and `stale_since_revision`. 2. `strands` contains ordered `top` and `bottom` lists of `Segment` objects. 3. The top strand runs 5′→3′; the bottom runs 3′→5′, aligned beneath the top. 4. An empty strand list represents an absent strand. 5. Each `Segment` has `name`, `type`, and `origin`; names describe symbolic regions, not base sequences. 6. Segment types: `adapter`, `barcode`, `umi`, `insert`, `primer`, `tso`, `handle`, `index`, `polyA`, `other`. 7. Origins: `source`, `skill`, `memory`, `llm`, `human`; `evidence` is a list of reference strings. 8. `skill_call_id` is a string or `None`; `stale_since_revision` is an integer or `None`. 9. Review status: `unreviewed`, `accepted`, `modified`, `rejected`, or `unresolved`. 10. Pydantic validates the schema strictly, forbids extra fields, and rejects nucleotide strings in symbolic fields. ## Install Requires Python 3.11 or newer. From this checkout: ```sh python3 -m venv .venv . .venv/bin/activate python -m pip install . ``` The distribution is `cdna-engine`; the import is `cdna`. Runtime dependencies are Pydantic and the MCP SDK. No database, model credentials, ground truth, or host application is needed. ## Two skills | Function / MCP tool | Substrate | Produc