Progressive Disclosure¶
Progressive disclosure is the primary interaction model in BioMCP.
The model is simple:
- search broadly,
- get a specific record,
- request only the section you need.
Why this pattern exists¶
Biomedical APIs can return large nested payloads. Sending full payloads by default wastes time and context space.
Progressive disclosure keeps default output concise and predictable.
Three-step workflow¶
Step 1: Search¶
Use search to identify candidates.
Step 2: Get¶
Use get to anchor on a specific identifier.
Step 3: Expand by section¶
Request a specific section only when needed.
Entity examples¶
Gene:
Variant:
Trial:
Drug:
Operational benefits¶
- Lower token usage for LLM-based workflows.
- Less irrelevant output in terminal usage.
- Faster debugging because each request has a narrow scope.
- Easier automation because command intent is explicit.
Practical guidance¶
- Start with
searchwhen you are unsure of identifiers. - Use one section at a time before requesting multiple sections.
- Keep
--jsonfor downstream systems that require structured parsing.
Relationship to MCP resources¶
MCP resources in BioMCP encode this same pattern in reusable workflows.
Use resources when you need guidance on sequence, and tools when you are ready to execute commands.
Summary¶
Progressive disclosure is not a cosmetic choice. It is the mechanism that keeps BioMCP usable for both humans and agents at scale.