BioMCP

# Progressive Disclosure Progressive disclosure is the primary interaction model in BioMCP. The model is simple: 1. search broadly, 2. get a specific record, 3. 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. ```bash biomcp search article -g BRAF -d melanoma --limit 5 ``` ### Step 2: Get Use `get` to anchor on a specific identifier. ```bash biomcp get article 22663011 ``` ### Step 3: Expand by section Request a specific section only when needed. ```bash biomcp get article 22663011 fulltext ``` ## Entity examples Gene: ```bash biomcp get gene BRAF biomcp get gene BRAF pathways ``` Variant: ```bash biomcp get variant "BRAF V600E" biomcp get variant "BRAF V600E" predict ``` Trial: ```bash biomcp get trial NCT02576665 biomcp get trial NCT02576665 eligibility ``` Drug: ```bash biomcp get drug carboplatin biomcp get drug carboplatin shortage ``` ## 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 `search` when you are unsure of identifiers. - Use one section at a time before requesting multiple sections. - Keep `--json` for 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.