Input modes
Repo, website, docs, owner notes, screenshots, mixed sources
Pre-Onboarding Kit
Give your AI coding agent one prompt. It turns the product material you actually have - repo, docs, website links, API specs, support exports, owner notes, policies, recordings, or screenshots - into an AnswerLattice-ready upload package.
Repo, website, docs, owner notes, screenshots, mixed sources
Reviewed source package, upload skeletons, support tests, asset rules
Prepares inputs; does not publish official answers
Choose your builder
Every package contains the same master safety and review rules. The wrapper only explains how to begin in that environment.
What it prepares
The prompt does the preparation work outside AnswerLattice, then leaves the owner with a reviewed package to upload.
Product context, workflows, support topics, website truth, policy boundaries, and source evidence maps.
Manifest and AnswerLattice add-source payloads that match the generated source files.
Owner-style questions that prove AnswerLattice can answer routine support and escalate risky topics.
Asset slots, demo walkthrough briefs, capture plan, scrub level, and approval gates before public website use.
What to prepare, how to run the prompt, how to review the generated folder, and when to enable live support.
OpenRules for the AI IDE: source-first inspection, missing-source handling, validation, and confidence language.
OpenA human-readable runbook for repo, website-only, docs-only, owner-notes, and early-product paths.
OpenWorkflow
The agent prepares the package. AnswerLattice still reviews drafts before anything becomes live support.
Run the master prompt in a repo, docs workspace, website brief, or owner-notes workspace.
Check website pages, docs, app routes, policies, support flows, and screenshots.
Generate source files, payloads, a product-page map, support tests, and asset rules.
Remove private data, fix inaccuracies, and approve what can enter AnswerLattice.
Use Knowledge Intake to create review drafts from the prepared source set.
Go live only after answers, product pages, widget context, and test questions pass.
Source modes
A full repo is useful, but not mandatory. The prompt keeps AnswerLattice inputs structured while marking unavailable sources instead of guessing.
Safety boundary
Pre-onboarding makes the first AnswerLattice intake cleaner, but it only covers sources the agent can actually read. AnswerLattice still requires source review, owner approval, and production checks before answering live users.
No secrets, tokens, cookies, or service accounts.
No payment details, raw logs, or private customer records.
No private-source processing without permission for the selected AI tool, and no public citation of private source material.
No legal, privacy, refund, or security claims without approved source wording.
No ticket, chat, macro, or repeated reply becomes approved truth without review.
No guarantee that every AI IDE can inspect every private repo, login-only app, website, recording, or file.
Unavailable sources must be marked pending instead of treated as covered.
No live support until sources, product pages, widget context, and test questions pass.
Start before onboarding
Use the prompt first, review the output, then upload the selected sources into AnswerLattice Knowledge Intake.