# narrative_writer

The narrative_writer prompt for Claude Code and ChatGPT builds a narrative section for a clinical-trial forecast report. It traces the model from phase 2 evidence to the estimated probability of a phase 3 outcome.

- prompt · `type=prompt` · Claude Code, ChatGPT · ★ 654 repo stars
- Author: anthropics · official Anthropic repository
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/anthropics/life-sciences/tree/HEAD/clinical-trial-forecast/prompts/narrative_writer.md
- Updated: 2026-11-09
- Security check: automated, 2026-11-17 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/prompts/narrative-writer

## What it does

The prompt structures the report section around the simulation stages, explaining each modelling choice and the evidence behind it. It calls for source tables, calculations, uncertainty estimates and conclusions about the outcome probability. Stages with little effect on the forecast can be combined; the report body has a 3,500-word limit, including tables.

The workflow draws on model files such as `sim/flow.json`, `sim/params.json`, `sim/parameter_metadata.json`, `sim/results.json` and `sim/summary_stats.json`, plus trial analyses and topic research. It instructs the model to preserve parameter values and outputs rather than recalculate them, use corrected figures from independent number checks, and include charts with specifications for additional visualisations. Web access is prohibited.

The prompt takes report details such as `report_kind`, `indication`, `asset`, audience and goal. It does not run the simulation itself. Installing it in Claude Code uses the marketplace commands and the `.claude/commands/narrative_writer.md` file.

## Install

### Claude Code (native platform)

1. **Plugin marketplace**
In the agent chat (not a shell):
```text
/plugin marketplace add anthropics/life-sciences
```
In the agent chat (not a shell):
```text
/plugin install clinical-trial-forecast@life-sciences
```

2. **Manually: put the files in place**
- File: `.claude/commands/narrative_writer.md`

### ChatGPT

**How to use**
- Paste the text at the start of a chat or into project instructions

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/anthropics/life-sciences/tree/HEAD/clinical-trial-forecast/prompts/narrative_writer.md

## When to use it
- explain how a clinical-trial forecast is built
- connect model assumptions to trial evidence
- show which stages contribute most to forecast uncertainty
- prepare report conclusions and phase 3 forecast figures

## Example prompts
- Write the forecast section for `{{indication}}` and `{{asset}}` using the model materials.
- Trace how the phase 2 evidence feeds into the estimated phase 3 outcome probability.
- Draft the simulation-stage narrative and show the main sources of forecast uncertainty.

## Similar
- [gmgn-narrative](https://affpapa.org/en/nevibecode/skills/gmgn-narrative.md) — skill · 7000 installs
- [research-paper-writer](https://affpapa.org/en/nevibecode/skills/research-paper-writer.md) — skill · 3300 installs
- [ccf-paper-writer](https://affpapa.org/en/nevibecode/skills/ccf-paper-writer.md) — skill · 2300 installs
- [ccf-rebuttal-writer](https://affpapa.org/en/nevibecode/skills/ccf-rebuttal-writer.md) — skill · 2200 installs
- [ai-paper-writer](https://affpapa.org/en/nevibecode/skills/ai-paper-writer.md) — skill · 1200 installs
- [paper-writer](https://affpapa.org/en/nevibecode/skills/paper-writer.md) — skill · 697 installs

From the same set: [anthropics/life-sciences](https://affpapa.org/en/nevibecode/by/anthropics/life-sciences.md) — 29 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
