Claude

Verified against Claude · 2026-08-07

Generate a role-specific onboarding doc from company knowledge without leaking what that role shouldn't see

A prompt for Claude Enterprise or Team that turns shared company knowledge into a new-hire onboarding document scoped to one role, with an access boundary enforced by omission rather than by mention, so restricted content never even gets hinted at.

Claude EnterpriseClaude TeamClaude (Projects)5 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

Produce an onboarding document for a new hire in the role below, drawn only from the company knowledge already available in this workspace — do not invent a process, tool, or policy that isn't actually documented, even a plausible-sounding one that would normally exist at a company like this.

ROLE BEING ONBOARDED
Mid-level Customer Success Manager, assigned to the Enterprise account segment

BACKGROUND OF THIS HIRE
This person has 4 years of CS experience elsewhere, so skip generic 'what is customer success' framing and go straight to how it works specifically here.

WHAT THIS ROLE ACTUALLY NEEDS TO KNOW IN THE FIRST TWO WEEKS
Our support escalation process and SLAs, the CS-to-Sales handoff for renewals, how account health scores are calculated, and which tools (and access requests) they'll need in week one. Depth matters more than breadth here — cover these topics thoroughly enough to actually act on, rather than skimming a longer list of topics shallowly.

ACCESS BOUNDARY — WHAT THIS ROLE MUST NOT SEE
Anything from the 'Compensation & Equity' or 'Layoff Planning' knowledge folders, even if a linked document references them in passing. If a source document you'd otherwise pull from also contains something in this excluded category, use only the parts that are in scope and do not include, summarize, or even vaguely allude to the excluded parts — an onboarding doc that hints at the existence of restricted information ("there's also a separate compensation framework, but that's not covered here") can be as much of a leak as including it outright, so state the boundary by omission, not by pointing at what's on the other side of it.

STRUCTURE
1. A short "what this role owns" section — the actual scope of responsibility, stated concretely, not a generic job-description paragraph already covered in the offer letter.
2. For each topic in scope, the current documented process or fact, with a note on which internal document it came from, so anything that goes stale can be traced back to its source and updated at the root rather than in this derived doc.
3. A "who to ask" section naming which team or role owns each topic, if that information exists in the available knowledge — do not invent a name or team if this isn't actually documented, say that the knowledge base doesn't specify an owner instead.
4. A "what's deliberately not covered here" closing note, naming topics that exist at the company but fall outside this role's first-two-weeks scope, without describing what they actually are — this manages expectations about the doc's boundaries without leaking access-restricted content through vague foreshadowing.

CONSTRAINTS
- If the available company knowledge has a real gap on a topic this role needs, say so explicitly in that section rather than filling the gap with a generic best-practice description that isn't actually this company's documented process.
- If two source documents disagree about a process this role needs, flag the conflict by name rather than picking one silently — an onboarding doc that resolves an internal disagreement invisibly hands the new hire confident wrong information with no way to know it was ever in question.
- Plain and direct, structured with headers, aiming for something a new hire can read start to finish in under 20 minutes.

OUTPUT
The complete onboarding document, structured as above, ready to hand to the new hire directly.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

Grounding strictly in available company knowledge, and forbidding plausible invention, matters because an onboarding doc is exactly the kind of generic-sounding content a model can produce convincingly from training-data patterns about how companies typically onboard a role like this one, and the actual risk is a new hire receiving a document that reads as authoritative company process but is really a well-written generic template with this company's role title swapped in; requiring every process claim to trace to an actual source document forces the output to be this company's real, current state rather than a plausible simulation of it. Boundary-by-omission, rather than boundary-by-mention, closes a subtler leak a naive access-control instruction still allows: acknowledging that restricted content exists and roughly where it lives is itself information a new hire with no other context didn't have before — there is a separate compensation framework, and it's kept apart from this — so stating the rule as 'omit without alluding to' closes a leak vector that a cruder exclusion instruction would let straight through. Requiring source attribution per topic turns the onboarding doc into a derived artifact traceable back to its root documents rather than a standalone thing that will itself go stale and become the only copy of outdated information floating around; when the underlying process changes, whoever maintains the knowledge base knows to update the source, and this doc's own claims can be checked against it later instead of being trusted as ground truth indefinitely. Flagging cross-document conflicts and knowledge gaps explicitly, instead of silently resolving them, is the single highest-risk failure mode for a synthesized onboarding document specifically, because a new hire has no independent way to know two internal sources disagreed and one was picked arbitrarily — naming the conflict, or naming the gap, hands the new hire an accurate picture of what is actually known and unsettled internally rather than false confidence dressed as a finished document.

Verified against

Claude Sonnet 4.6 (Enterprise) · 2026-08-07

Changelog

  • 2026-08-07 — Initial publish, verified against Claude Sonnet 4.6 in a Claude Enterprise workspace.

Building this for real?

This is a free starting point. If you'd rather have what Scult builds built and running for your business, that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
Set up a Claude Project that stays accurate for months, not just this chatA structured Claude Projects setup that splits durable knowledge from custom instructions and adds an explicit staleness-check protocol, so every new chat in the Project inherits accurate context automatically instead of the knowledge base quietly rotting into confidently wrong answers.Claude (Projects)Claude Enterprise2026-07-20Build a multi-view dashboard as one self-contained Claude ArtifactA prompt for building a single-file interactive dashboard Artifact with multiple tabs sharing one dataset and one state model, so filters and selections stay consistent across views instead of several disconnected mini-tools bolted together.Claude (Artifacts)Claude.ai2026-07-21Audit a full contract for risk using Claude's whole context window, not a skimA long-context prompt that forces a clause-by-clause, quote-grounded risk audit of an entire pasted contract or policy document, ranked by severity and tied to one party's actual position, instead of a general summary of what the document is about.Claude (Opus 4.6, 1M context)Claude (Sonnet 4.6, 200K context)2026-07-22Feed Claude untrusted or user-submitted text without letting it hijack your instructionsAn XML-tagged prompt structure for any task where Claude must process content you did not write yourself — customer submissions, scraped pages, forum posts — that keeps instructions and untrusted data in separate labeled blocks and requires flagging any embedded instruction-injection attempt.Claude (Sonnet 4.6)Claude (Opus 4.6)2026-07-23
All Claude prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY