Skip to main content
prompt GP ChatGPT official · Anthropic

condense_part — Claude Code prompt for report sections

condense_part is a Claude Code prompt that trims a clinical forecast report section to a word limit. It keeps key figures, tables, and citations intact.

no license ★ 654 updated 2026-11-09

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add anthropics/life-sciences
/plugin install clinical-trial-forecast@life-sciences
Open on GitHub

What condense_part does

condense_part reduces one section of a forecast report for a given indication and asset to a hard word ceiling, tables included. The template takes the report type, indication, asset, word limit, current word count, minimum required cut, and a description of uncertainty drivers.

It allocates cuts by how much each stage affects the estimate. Low-impact stages are grouped, while repeated material and secondary details are removed. The prompt preserves headline estimates and intervals, key tables and figures, model-choice rationale, closing conclusions, and appendix equations. Kept figures, citations, numbers, and parameter names must remain unchanged.

The prompt has no tools and cannot add facts. Its output must contain only the section in Markdown; if it remains over the limit, it allows further passes, up to three.

What it helps with

  • cut a forecast section to a word ceiling
  • retain key estimates and intervals
  • group stages with little effect on the estimate
  • remove repeated tables and background

Example prompts for the agent

  • Condense this report section to 800 words and keep the key tables and citations.
  • Trim the forecast section for {{indication}} while preserving headline estimates and intervals.
  • Run /condense_part on this section with a 600-word ceiling.

How to install condense_part

1Plugin marketplace

/plugin marketplace add anthropics/life-sciences
/plugin install clinical-trial-forecast@life-sciences

2Manually: put the files in place

.claude/commands/condense_part.md

Questions about condense_part

How do I install condense_part in Claude Code?

Run /plugin marketplace add anthropics/life-sciences, then /plugin install clinical-trial-forecast@life-sciences. The command file is .claude/commands/condense_part.md.

What does condense_part do?

It cuts a forecast report section to a specified word limit, prioritizing material by its effect on the estimate. It retains key estimates, supporting tables and figures, citations, and appendix equations.

What information does condense_part need?

Its template takes the report type, indication, asset, word ceiling, current word count, required cut, and the factors that carry uncertainty in the forecast.

Similar prompts

All prompts for Science and research →

From the same set29

The whole set →
forecast_scope The forecast_scope Claude Code prompt sets up a probability-of-technical-success forecast for a registered clinical trial. It turns the request and registry record into a project.json configuration. GP ChatGPT ★ 654repo stars followup_triage The followup_triage prompt for Claude Code and ChatGPT sorts researchers’ follow-up questions for another research round. It removes duplicates and questions already answered. GP ChatGPT ★ 654repo stars copyedit_section copyedit_section — a Claude Code prompt revises clinical-trial forecast sections. It reshapes dense prose while retaining figures, citations, and document structure. GP ChatGPT ★ 654repo stars condense_trials The condense_trials prompt for Claude Code turns clinical-trial cards into a single digest. It groups programs by mechanism and phase, with findings tied to the report goal. GP ChatGPT ★ 654repo stars classify_trial The classify_trial Claude Code prompt classifies clinical trial records for a specified indication. It takes one registry entry and returns data in a JSON schema. GP ChatGPT ★ 654repo stars card_checker The card_checker Claude Code prompt checks key figures against the sources cited in an answer. It flags mismatches in doses, study arms, timepoints, units, and populations. GP ChatGPT ★ 654repo stars apply_ledger The apply_ledger Claude Code prompt applies independent fact-check findings to a report section. It revises claims flagged in the ledger while preserving untouched content. GP ChatGPT ★ 654repo stars appendix_writer appendix_writer — a prompt for Claude Code and ChatGPT that compiles technical simulation appendices for clinical trials. It converts model files into structured audit documentation without external web calls. GP ChatGPT ★ 654repo stars

Project sponsors

Companies that keep this analytics open