DocsIntelligence ModulesOnboarding Map
🚀

Onboarding Map

A reading guide for your repository, written from your repository. Fourteen sections take a new engineer from “what is this system” to “here is how you make your first change” — and each one says where its claims came from, so nothing an AI guessed can pass as a measured fact.

How it is built

The guide is generated from the Intent Map, the dependency graph and the findings of the other scans. It is not a separate analysis of your code — it is a reading of the analyses you have already run, which is why it regenerates in seconds once the Intent Map exists.

NOTE
Generate the Intent Map first. Without file summaries the guide has nothing to read, and the tab says so rather than producing an empty one.

The tab

Sections that have nothing to show collapse themselves, and the rail on the right lists only the ones that rendered.

mapnostics.com/repo/mapnostics-app/onboarding
Generated 31 Aug 2026, 09:31Re-generate

Partial scan — 296 of 312 indexed code files summarised. Re-run the Intent Map to complete the scan, then re-generate this guide.

Architecture OverviewInference

A Next.js App Router application over a Postgres database with pgvector. Repository content is indexed into chunks, then a set of independent scans read those chunks and write findings that the tabs render.

Next.jsTypeScriptSupabaseTailwindVercel
Repository at a GlanceVerified
312files
48directories
41entry points
5technologies
~9hcore path
Key Entry PointsVerified
API routeapp/api/repos/[id]/dead-code/route.tsRuns the batched dead-code pipeline · high coupling
Middlewaremiddleware.tsSession refresh on every request
Cronapp/api/cron/autopilot/route.tsDaily re-index at 09:00 UTC
Suggested Learning PathEstimated
1
Read how a repository is indexed~2h
lib/pipeline/index-repo.ts, app/api/repos/[id]/index/route.ts
2
Follow one scan end to end~3h
lib/pipeline/detect-dead-code.ts, components/tabs/dead-code-tab.tsx
3
Learn the access gate every read passes~1h
lib/supabase/repo-access.ts
Deliberately Out of Scope

This guide is built entirely from your code and its history. Some things a new engineer needs are decisions people made, so Mapnostics does not guess at them:

Business context & goalswhy the product exists and what it is optimising for
Revenue & billing flowshow money moves through the system
Team ownershipwhich team or person owns each area
Escalation & on-call pathswho to page when something breaks in production

The guide sections

Orientation

  • Coverage — a banner when fewer files were summarised than indexed, naming the gap. It disappears entirely on a complete scan.
  • Architecture Overview — what the system is, in prose, with the technologies detected in it.
  • Repository at a Glance — files, directories, entry points, technologies, and the estimated hours for the core path.
  • Runtime Flow — how a request moves through the system, derived from static analysis rather than from traces.

The map

  • Repository Tour — what each top-level directory is for, and the key files inside it.
  • Complexity by Area — where the coupling is concentrated, so you know which parts will fight back.
  • Key Entry Points — every API route, page, layout, middleware, webhook, cron job and CLI script, with its coupling and why it matters.
  • Watch Out For — the gotchas: the places where the obvious reading of the code is the wrong one.
  • High Risk Areas — the debt already found by the other scans — dead code, zombies, unused dependencies and assets, orphan tables, secrets — each linking to the tab that found it.

The path

  • Suggested Learning Path — numbered steps, each with the files to read, an hours estimate, and why the step matters. Ordered by dependency, not by file name.
  • Learning Path by Role — the same repository read differently for a frontend, backend or full-stack starting point.
  • Change Recipes — how to do the five common changes in this codebase: add an endpoint, add a database column, add a background job, add a feature flag, add a webhook. Each step cites the evidence behind it, and recipes this repository has no surface for are named as not applicable rather than faked.
  • Next Recommended Analyses — which scans have not run yet, so the gaps in the guide are visible.
NOTE
High Risk Areas distinguishes “this audit found nothing” from “this audit has never run”. A source that has not run is listed as unavailable, never rendered as clean.

Provenance badges

Every section carries a badge naming where its content came from and how far it can be trusted:

  • Verified — derived from code: the AST, the import graph, static analysis. High trust.
  • Provided — taken from your repository configuration or documentation. High trust.
  • Estimated — a statistical heuristic, such as the hours on a learning step. Medium trust.
  • Inference — an AI interpretation. A starting point, not a fact. Low trust.

The Architecture Overview is the one low-trust header on the page, and it wears the badge rather than being dressed up as fact. Each badge also names the concrete sources behind its section and the limits of what those sources cover.

NOTE
There is an Observed level in the taxonomy, reserved for real runtime traces. Nothing in Mapnostics uses it today, because nothing in Mapnostics watches your code run. The Runtime Flow section is static analysis and is badged Verified, not Observed.

What it will not guess

The last section of the guide is a deliberate boundary. Business context, revenue flows, team ownership and escalation paths are decisions people made, not facts in the code. Mapnostics names them as out of scope rather than leaving a gap that reads like a missing feature — or worse, filling it with a confident guess.

TIP
Point a new engineer at the Suggested Learning Path, and tell them to run the Impact Simulator before their first change. The two together turn “I do not know what this touches” into an answer they can check.

See Developer Onboarding on the product page →