# anthropics/life-sciences architecture

> Architecture diagram of the GitHub repository anthropics/life-sciences, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-17). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/anthropics/life-sciences
- Repository: https://github.com/anthropics/life-sciences

## Overview

This repository is a Claude Code marketplace for life-sciences skills, not one integrated runtime application. Its main analytical workflow is Clinical Trial Forecast: a session workflow coordinates deterministic tools and model/report stages to produce an evidence-backed probability and report. Separate skills cover protocol generation, scientific problem selection, and binder validation. The README documents their capabilities, but their detailed orchestration is mostly in unsampled skill files; those branches are therefore represented at the skill-directory level. Forecast edges reflect only visible imports, workflow calls, or documented capabilities.

Main entry points: [`clinical-trial-forecast/scripts/ft_tools.py`](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_tools.py) (Tool command router), [`README.md`](https://github.com/anthropics/life-sciences/blob/main/README.md) (Claude Code marketplace), [`binder-gauntlet/SKILL.md`](https://github.com/anthropics/life-sciences/blob/main/binder-gauntlet/SKILL.md) (Binder Gauntlet), [`clinical-trial-protocol-skill/SKILL.md`](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-protocol-skill/SKILL.md) (Protocol generation)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_marketplace["Marketplace skills"]
  node_market["Claude Code marketplace<br/>[README.md]"]
  node_problem["Problem selection"]
end

subgraph group_forecast["Trial forecasting"]
  node_forecast_skill["Forecast skill<br/>[SKILL.md]"]
  node_workflow["Session workflow<br/>[cs_workflow.py]"]
  node_tools["Tool command router<br/>[ft_tools.py]"]
  node_base["Shared data and prompts<br/>[ft_base.py]"]
  node_registry["Trial registry<br/>[ft_registry.py]"]
  node_research["Trial research<br/>[ft_research.py]"]
  node_model["Forecast simulation<br/>[ft_model.py]"]
  node_report["Report and claims<br/>[ft_report.py]"]
  node_figures["Report figures<br/>[ft_figures.py]"]
  node_data_pack["Local data packing<br/>[add_data.py]"]
end

subgraph group_protocol["Protocol design"]
  node_protocol_skill["Protocol generation<br/>[SKILL.md]"]
  node_protocol_steps["Protocol workflow"]
  node_sample_size["Sample-size calculator"]
end

subgraph group_binder["Binder validation"]
  node_binder_skill["Binder Gauntlet<br/>[SKILL.md]"]
  node_sequence_profile["Sequence profiling"]
  node_binder_controls["Prediction and controls"]
  node_binder_dossier["Developability dossier"]
end

node_researcher(("Life-sciences researcher"))
node_trial_registry_service["ClinicalTrials.gov"]

node_researcher -->|"chooses skills"| node_market
node_market -->|"offers"| node_forecast_skill
node_market -->|"offers"| node_protocol_skill
node_market -->|"offers"| node_problem
node_market -->|"offers"| node_binder_skill
node_researcher -->|"requests forecast"| node_forecast_skill
node_forecast_skill -.->|"directs session"| node_workflow
node_workflow -->|"runs commands"| node_tools
node_tools -->|"imports helpers"| node_base
node_tools -->|"imports commands"| node_registry
node_tools -->|"imports commands"| node_research
node_tools -->|"imports commands"| node_model
node_tools -->|"imports commands"| node_report
node_workflow -.->|"uses registry data"| node_trial_registry_service
node_researcher -->|"requests protocol"| node_protocol_skill
node_protocol_skill -.->|"follows workflow"| node_protocol_steps
node_protocol_skill -.->|"calculates sample size"| node_sample_size
node_researcher -->|"seeks strategy"| node_problem
node_researcher -->|"submits designs"| node_binder_skill
node_binder_skill -->|"profiles sequences"| node_sequence_profile
node_binder_skill -.->|"validates with controls"| node_binder_controls
node_binder_skill -.->|"writes dossier"| node_binder_dossier

click node_market "https://github.com/anthropics/life-sciences/blob/main/README.md"
click node_problem "https://github.com/anthropics/life-sciences/tree/main/scientific-problem-selection"
click node_forecast_skill "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/SKILL.md"
click node_workflow "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/cs_workflow.py"
click node_tools "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_tools.py"
click node_base "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_base.py"
click node_registry "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_registry.py"
click node_research "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_research.py"
click node_model "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_model.py"
click node_report "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_report.py"
click node_figures "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_figures.py"
click node_data_pack "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/add_data.py"
click node_protocol_skill "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-protocol-skill/SKILL.md"
click node_protocol_steps "https://github.com/anthropics/life-sciences/tree/main/clinical-trial-protocol-skill/references"
click node_sample_size "https://github.com/anthropics/life-sciences/blob/main/clinical-trial-protocol-skill/scripts/sample_size_calculator.py"
click node_binder_skill "https://github.com/anthropics/life-sciences/blob/main/binder-gauntlet/SKILL.md"
click node_sequence_profile "https://github.com/anthropics/life-sciences/blob/main/binder-gauntlet/scripts/sequence_profile.py"
click node_binder_controls "https://github.com/anthropics/life-sciences/tree/main/binder-gauntlet"
click node_binder_dossier "https://github.com/anthropics/life-sciences/tree/main/binder-gauntlet"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_market,node_problem toneBlue
class node_forecast_skill,node_workflow,node_tools,node_base,node_registry,node_research,node_model,node_report,node_figures,node_data_pack toneAmber
class node_protocol_skill,node_protocol_steps,node_sample_size toneMint
class node_binder_skill,node_sequence_profile,node_binder_controls,node_binder_dossier toneRose
class node_researcher,node_trial_registry_service toneIndigo
```

## Components

### Marketplace skills

- [Claude Code marketplace](https://github.com/anthropics/life-sciences/blob/main/README.md): `README.md`
- [Problem selection](https://github.com/anthropics/life-sciences/tree/main/scientific-problem-selection): `scientific-problem-selection`

### Trial forecasting

- [Forecast skill](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/SKILL.md): `clinical-trial-forecast/SKILL.md`
- [Session workflow](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/cs_workflow.py): `clinical-trial-forecast/scripts/cs_workflow.py`
- [Tool command router](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_tools.py): `clinical-trial-forecast/scripts/ft_tools.py`
- [Shared data and prompts](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_base.py): `clinical-trial-forecast/scripts/ft_base.py`
- [Trial registry](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_registry.py): `clinical-trial-forecast/scripts/ft_registry.py`
- [Trial research](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_research.py): `clinical-trial-forecast/scripts/ft_research.py`
- [Forecast simulation](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_model.py): `clinical-trial-forecast/scripts/ft_model.py`
- [Report and claims](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_report.py): `clinical-trial-forecast/scripts/ft_report.py`
- [Report figures](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/ft_figures.py): `clinical-trial-forecast/scripts/ft_figures.py`
- [Local data packing](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-forecast/scripts/add_data.py): `clinical-trial-forecast/scripts/add_data.py`

### Protocol design

- [Protocol generation](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-protocol-skill/SKILL.md): `clinical-trial-protocol-skill/SKILL.md`
- [Protocol workflow](https://github.com/anthropics/life-sciences/tree/main/clinical-trial-protocol-skill/references): `clinical-trial-protocol-skill/references`
- [Sample-size calculator](https://github.com/anthropics/life-sciences/blob/main/clinical-trial-protocol-skill/scripts/sample_size_calculator.py): `clinical-trial-protocol-skill/scripts/sample_size_calculator.py`

### Binder validation

- [Binder Gauntlet](https://github.com/anthropics/life-sciences/blob/main/binder-gauntlet/SKILL.md): `binder-gauntlet/SKILL.md`
- [Sequence profiling](https://github.com/anthropics/life-sciences/blob/main/binder-gauntlet/scripts/sequence_profile.py): `binder-gauntlet/scripts/sequence_profile.py`
- [Prediction and controls](https://github.com/anthropics/life-sciences/tree/main/binder-gauntlet): `binder-gauntlet`
- [Developability dossier](https://github.com/anthropics/life-sciences/tree/main/binder-gauntlet): `binder-gauntlet`

### Other components

- **Life-sciences researcher**
- **ClinicalTrials.gov**

## Connections

- Life-sciences researcher → Claude Code marketplace: chooses skills (evidence: `README.md`)
- Claude Code marketplace → Forecast skill: offers (evidence: `README.md`)
- Claude Code marketplace → Protocol generation: offers (evidence: `README.md`)
- Claude Code marketplace → Problem selection: offers (evidence: `README.md`)
- Claude Code marketplace → Binder Gauntlet: offers (evidence: `README.md`)
- Life-sciences researcher → Forecast skill: requests forecast (evidence: `README.md`)
- Forecast skill → Session workflow: directs session
- Session workflow → Tool command router: runs commands (evidence: `clinical-trial-forecast/scripts/cs_workflow.py`)
- Tool command router → Shared data and prompts: imports helpers (evidence: `clinical-trial-forecast/scripts/ft_tools.py`)
- Tool command router → Trial registry: imports commands (evidence: `clinical-trial-forecast/scripts/ft_tools.py`)
- Tool command router → Trial research: imports commands (evidence: `clinical-trial-forecast/scripts/ft_tools.py`)
- Tool command router → Forecast simulation: imports commands (evidence: `clinical-trial-forecast/scripts/ft_tools.py`)
- Tool command router → Report and claims: imports commands (evidence: `clinical-trial-forecast/scripts/ft_tools.py`)
- Session workflow → ClinicalTrials.gov: uses registry data
- Life-sciences researcher → Protocol generation: requests protocol (evidence: `README.md`)
- Protocol generation → Protocol workflow: follows workflow
- Protocol generation → Sample-size calculator: calculates sample size
- Life-sciences researcher → Problem selection: seeks strategy (evidence: `README.md`)
- Life-sciences researcher → Binder Gauntlet: submits designs (evidence: `README.md`)
- Binder Gauntlet → Sequence profiling: profiles sequences (evidence: `binder-gauntlet/scripts/sequence_profile.py`)
- Binder Gauntlet → Prediction and controls: validates with controls
- Binder Gauntlet → Developability dossier: writes dossier
