Dipio Evidence Gate

Read the research. Stand on the evidence. Build.

The Dipio Evidence Gate is the entry point for coding agents into the Dipio Evidence Loop. Your agent connects through the Dipio MCP and reads the same user research your team does: qualified interview counts, reports with traceable evidence, implementation-ready specs and tickets, and a way to draft the next study when the evidence runs out.

Every finding is traceable to the conversation it came from, so the code your agent ships is backed by the research it stands on.

Six tools.

The full surface your coding agent sees through the Gate. Read access to findings, specs, and tickets, plus a drafting tool for what the research does not yet cover.

list_studies()

Every study with its qualified interview count. Only interviews above a duration floor and with a passing quality score are counted, so the number you see is the number you can rely on.

get_report(study_id)

Full findings with a rigor block: evidence_n (the count the findings rest on, and the number to report), the live qualified_n, a staleness flag, and verbatim quotes with their conversation ids.

get_spec(study_id, format?)

Implementation-ready feature specs from behaviour-diagnostics studies: requirements from the screened techniques, success criteria from the experiment design, and cautions from the screening. Other methodologies get an honest, structured refusal instead of a guess.

get_ticket(study_id)

Evidence-backed tracker tickets for any study, ready to drop into your issue tracker with the research they stand on attached.

draft_study(...)

When the research does not answer a question, the agent drafts the study that would answer it. A human reviews and publishes. Drafts are inert: nothing is collected or billed until publish.

research_context(question)

Ask anything. Get report insights plus verbatim interview excerpts with provenance, or an explicit evidence_gap when the research does not cover it, never an invented answer.

Connect any MCP client.

The Evidence Gate works with the coding agents and MCP clients your team already uses. The connection config lives in the app: no API keys to paste into this page. It is included with every plan, with no per-call metering.

Claude CodeCursorConductorKiroCodex CLIany MCP client

Find the config in Settings → Agents after signup.