ask/engine

Introduction

What AskEngine is and how your agent reads it.

AskEngine interviews your users and turns what they say into a belief layer: a living, queryable model of what your users need and how confident you can be about it. Every belief traces back to the exact interview moments that justify it.

The front door is your coding agent. Connect the MCP server once with your team API key (from team settings), and Claude Code, Cursor, or Codex can answer questions like "what do we currently believe about onboarding?" — with citations.

claude mcp add --transport http askengine https://askengine.dev/api/mcp \  --header "Authorization: Bearer YOUR_API_KEY"

Then ask your agent something. It discovers the tools on its own — including ask, which hands a whole question to the belief layer and returns one answer with verified citations. The same surface is also a REST API for scripts, dashboards, and cron jobs.

The chain

Everything in AskEngine is one chain: studies collect interviews, interviews produce evidence, evidence supports signals, and signals drive recommendations — while research gaps track what you don't know yet, and the inbox tells you what needs a human. Each link has its own page in the Platform section.

Four objects carry the reasoning. The fastest way to tell them apart is by contrast:

ThemeSignalRecommendationResearch gap
What it isA pattern inside one studyA durable belief held across studiesA concrete action derived from signalsA named unknown
Question it answers"What did this study find?""What does the team believe?""What should we do about it?""What don't we know yet?"
ScopeOne studyThe whole teamThe whole teamOne signal
How it closesMerged or archived with its studyArchived: validated or abandonedAn outcome is recordedApplied to a study, or dismissed

Where next

  • Platform — one short page per object: what it is, when to trust it, how to correct it.
  • MCP — connect any client, plus the full tool table.
  • API — the REST surface behind everything, with an OpenAPI 3.1 spec.