7 added, 7 removed. Audit A to A.
---
name: indication-dossier
description: >
Generate a therapeutic indication dossier. Covers the patient population,
epidemiology, disease biology, standard of care, regulatory precedent, and
landmark clinical trials.
license: Apache-2.0
---
# Indication Dossier
Produces a structured research dossier on a single indication, framed as a
patient population: who they are, what's wrong, how they're treated today,
and how clinical trials can be designed to help them. Runs as five phases
that write resumable waypoint files; after a brief identity check at the end
of Phase 1, the remaining phases run straight through.
## Framing
**Think of an indication as a patient population.** Frame everything
from the patient perspective: "Who are these patients?" not "What is this
disease?"; "How are these patients identified and managed?" not "What causes
this condition?"; population nesting: "all patients in {child} are patients
in {parent}".
Some indications don't map to ICD codes or standard disease definitions:
"immunosenescence" is a biological state, not a billable diagnosis; "ageing"
is not an FDA-accepted indication; "GLP-1 induced sarcopenia" is an
iatrogenic population. Note these distinctions explicitly. They matter for
regulatory path and trial design.
## Inputs
- **`indication`** (required) — indication name (e.g., "sarcopenia",
"idiopathic pulmonary fibrosis").
- **`additional_context`** (optional) — areas to focus on, parent
indication, or other framing.
- **`workdir`** (optional) — where to write waypoints and the final report.
Defaults to `./do_not_commit/indication-dossier-<slug>/`.
## Tools this skill expects
- | Purpose | Tool |
- |---|---|
- | ClinicalTrials.gov | `clinical-trials` MCP |
- | Literature | `pubmed` MCP |
- | Web | `WebSearch`, `WebFetch` — FDA guidance, treatment guidelines (NCCN, AASLD, specialty societies), CDC/WHO epidemiology data |
- | Documents | `WebFetch` for remote PDFs; `Read` for local PDFs |
- | Subagents | `Agent` for parallel evidence gathering |
+ | Purpose | Tool |
+ | ------------------ | -------------------------------------------------------------------------------------------------------------------------- |
+ | ClinicalTrials.gov | `clinical-trials` MCP |
+ | Literature | `pubmed` MCP |
+ | Web | `WebSearch`, `WebFetch` — FDA guidance, treatment guidelines (NCCN, AASLD, specialty societies), CDC/WHO epidemiology data |
+ | Documents | `WebFetch` for remote PDFs; `Read` for local PDFs |
+ | Subagents | `Agent` for parallel evidence gathering |
If a listed MCP isn't connected, say so and fall back to `WebSearch` against
the underlying public source (clinicaltrials.gov, pubmed.ncbi.nlm.nih.gov).
## Output layout
```
<workdir>/
└── waypoints/
├── progress.json # loop control
├── meta.json # phase 1
├── epidemiology.json # phase 2
├── biology_soc.json # phase 3
├── regulatory_trials.json # phase 4
├── sources_evaluated.json
├── research_output.json # phase 5 — structured output
└── indication_dossier_report.md # phase 5 — the deliverable
```
Schemas for every waypoint file are in `references/waypoint-schemas.md`.
Waypoints are the resumable state. If the workdir already has waypoints, read
them, summarize what's done, and ask which phase to resume from.
## Before starting
Read `references/00-research-standards.md`. It governs sourcing and the
anti-fabrication rules for every phase. Then create `<workdir>/waypoints/`.
## Workflow
The dossier is built in five phases. After each phase, write the waypoint
file and emit a ≤200-word summary of what you found and what's uncertain,
then proceed directly to the next phase. The one exception is Phase 1: after
writing `meta.json`, show the resolved indication identity and call
`ask_user` with options **Proceed** / **Revise identity** / **Stop**, so a
misread indication name can be caught before the expensive phases run. If
`ask_user` is unavailable, state "proceeding on this interpretation;
interrupt now to correct it" and continue.
### Phase 1 — Meta initialization
Read `references/01-meta-initialization.md`. Resolve the indication identity:
clinical definition, ICD codes, aliases, parent indication, and whether it's
a recognized diagnostic entity. Run a quick CT.gov landscape scan. Stand up
`waypoints/meta.json`.
### Phase 2 — Epidemiology research
Read `references/02-epidemiology-research.md`. Characterize the population:
diagnostic criteria, prevalence and incidence, demographics and risk factors,
natural history. Use parallel subagents to search PubMed and the web
simultaneously. Write `waypoints/epidemiology.json`.
### Phase 3 — Biology & standard-of-care research
Read `references/03-biology-soc-research.md`. Establish pathophysiology,
biomarkers, approved therapies, treatment guidelines, and unmet need. Use
parallel subagents: PubMed for biology, web for guidelines, FDA for
approvals. Write `waypoints/biology_soc.json`.
### Phase 4 — Regulatory & trials research
Read `references/04-regulatory-trials-research.md`. Establish FDA/EMA
accepted endpoints, regulatory precedents, typical trial design parameters,
landmark trials, and notable failures. Use parallel subagents: FDA for
guidance/approvals, CT.gov for trial patterns, PubMed for trial-history
reviews. Write `waypoints/regulatory_trials.json`.
### Phase 5 — Synthesis
Read `references/05-synthesis.md` and `references/06-writing-style.md`. Read
all four consolidated waypoint files. Write
`waypoints/indication_dossier_report.md` — narrative sections in the order
the synthesis reference specifies, with inline citations per the style guide
— and `waypoints/research_output.json`. No new research threads in this
phase. Targeted gap-fills are allowed: a single fetch to resolve a specific
missing value in an existing waypoint field (an approval year, an NCT ID, a
figure from a sponsor pipeline page). Anything broader than that, name as a
gap rather than filling it.
## Resuming
If invoked with a `workdir` that already contains waypoints: list which phases
are complete (waypoint file exists and is non-empty), show the meta summary,
and ask the user which phase to run next. Never overwrite an existing waypoint
without confirmation.