Gosset CLI
The Gosset CLI queries the Gosset database from your terminal: drugs, clinical trials, companies, deals, news and conference abstracts, plus two trial models, progression (PTRS) and timeline. Output is JSON by default, so it works with jq, pandas, spreadsheets or an AI agent.
It ships in the same package as the Python SDK: pip install gosset gives you both the gosset command and import gosset.
Why the CLI
- You type names, not ids. Pass a target, disease, drug, company or modality by name and the server resolves it.
--target PD-1,--disease "atopic dermatitis",--drug pembrolizumaball work. A Gosset id works too. - JSON by default. Pipe it into
jq, hand it to an agent, or add--tableto read it yourself. - One command per dataset.
drugs,trials,companies,deals,newsandabstractsshare the same paging, output and filter flags.
60-second tour
pip install gosset
gosset auth # one-time browser sign-in; the key is stored
# The TL1A landscape, most advanced first
gosset drugs --target TL1A --table
# Phase 3 semaglutide trials
gosset trials semaglutide --phase 3 --table
# TL1A deals, newest first
gosset deals --target TL1A --table
# Every abstract that reported on the DESTINY-Breast09 trial, at any meeting
gosset abstracts --nct NCT04784715 --table
# Names only, ready for the next step in a script
gosset drugs --target TL1A --phase 3 | jq -r '.[].name'Real output of gosset drugs --target TL1A --fields name,phase,developers --table --limit 5:
name phase developers
----------------------- ----- --------------------------------------
afimkibart 3 Roche, Roivant Sciences, Pfizer
duvakitug 3 Teva Pharmaceutical Industries, Sanofi
tulisokibart 3 Merck
risankizumab + ABBV-701 2 AbbVie
ABBV-701 2 AbbVie
… showing 5 of 75 totalAll commands
| Command | Use it for |
|---|---|
drugs | Assets by name, target, disease, modality, developer, phase |
trials | Trials by drug, disease, target, sponsor, phase, status, registry, dates |
companies | Companies by name, pipeline focus, country, type, stage |
deals | Licensing, M&A and financing by drug, target, disease, party, type, value, date |
news | News and press releases by text, drug, disease, company, category, date |
abstracts | Conference abstracts by drug, meeting, year, trial, target, impact |
ptrs, timeline | Progression estimate and predicted completion date for one trial |
schema | The field names each dataset returns |
usage | How much of your plan you have used |
skills | Ready-made workflows for AI agents |
auth, get-token, version | Sign-in and housekeeping |
The command reference lists every flag and default.
How names become filters
| You pass | The server | What it matched (live, November 2026) |
|---|---|---|
--target TL1A | finds the target and its ontology class | TNF-like cytokine 1A, 18 target records |
--disease "breast cancer" | classifies the disease and includes narrower diseases | breast cancer and its sub-classes |
--drug Keytruda | matches the drug's names, brands and codes | pembrolizumab (so does MK-3475) |
--company AstraZeneca | finds the company and its group records | AstraZeneca, 18 company records |
--target 680389902c09a077cc7ea8da | uses the Gosset id as given | one target record |
Add --debug to see the query the CLI sent. The JSON response from the API also says what each name matched; see Query API.
A name that matches nothing never widens the search
If a name can't be resolved (a typo, or not in the database), you get an empty result, a message on stderr and exit code 2. The filter is never dropped.
$ gosset drugs --target notarealtarget
[]
No match: could not resolve 'notarealtarget' (target). Check spelling, or pass a Gosset id.Where to next
- Getting started: install, sign in, output formats, paging
- drugs · trials · companies · deals · news · abstracts
- Predictions:
ptrsandtimeline - Query API and
--where: conditions the flags don't cover - Recipes: landscapes, competitor scans, deal comps, agent pipelines