Onboarding without the doc graveyard.

The problem​

Onboarding into a new team meant inheriting a folder of documents whose authors had largely left. Half the docs were stale; a third were contradictory; the rest were correct but assumed shared context the new joiner did not yet have. The team's onboarding ramp was eleven days, most of which was spent learning what to ignore.

Worse, the team had no way to communicate what they were collectively unsure of. A document either said something or it didn't. There was no place to write "we believe this, with this confidence, contradicting last year's claim, and we expect to revisit it." Known unknowns had no medium.

The approach — the scope is the team's brain​

The team treats project scope as the primitive a new joiner inherits. Switching scope is the onboarding action: the platform binds the joiner's agent to scope_helios, and the validated codex of project_helios becomes their working context — including its gaps, its conflicts, and its recent revisions.

Three commitments shape the experience:

• Scope inheritance is mechanical. A scope flip changes which gnoses an agent retrieves, which ethos governs the call, and which revision history surfaces. There is no manual context transfer.
• Known unknowns are gnoses too. When the team has explicitly chosen not to claim something, that gap is captured as a gnosis with a gap flag. New joiners see it.
• Recent revisions surface on day one. The first onboarding query returns not just the validated codex but also the gnoses revised in the last thirty days. New joiners learn what is changing, not just what is true.

A new engineer joins on a Monday and is asking sharp questions on Tuesday. We used to lose two weeks. Now we lose a couple of hours.

DFDiego F. — Engineering manager, project_helios

Shape of a scope inheritance​

A scope is a typed object: a name, a pinned ethos, a body of gnoses, an active revision pair, and a window of recent edits. Inheriting it is one operation that mints an op_id, binds the agent's context, and records the joiner's acknowledgement.

The codex itself is unchanged by the inheritance. The new joiner's agent now retrieves under the team's mandate; the team's prior agents continue to operate as they were. The handoff is a configuration event, not a content migration.

Operations & telemetry​

The scope-flip operation carries the source scope, the destination scope, the gnoses inherited, the gaps flagged, and the recent revisions surfaced. The record is the joiner's day-one trail.

Outcomes after one rotation cycle​

• 2
  Days median new-joiner ramp, down from eleven.

  New engineers stopped opening every doc in the project folder; they queried the scope and watched the platform return what was contested, recent, and validated.

• 100%
  Of known-unknowns surfaced on day one.

  Gaps are gnoses too. The platform reports what the team has explicitly chosen not to claim — which is often the most important page in the manual.

• −58%
  Repeated onboarding questions in team channels.

  The most-asked questions migrated into the codex as gnoses. New joiners now ask the agent first; the channel sees the residue, not the volume.

• 31
  Active scopes in production, each team-aligned.

  Scopes mirror team boundaries. Joining a team is now a scope flip; archiving a project is a scope freeze. The same primitive does both.

Implementation playbook​

Four decisions worth making early when you align scope with team boundaries.

1 · Define one scope per team, not per project.​

Projects come and go; teams persist longer. A team-aligned scope outlives any single project and accumulates the team's working knowledge across rotations.

2 · Capture gaps as gnoses, not as TODO comments.​

A gap gnosis with explicit references is far more useful than a "??" in a doc. The platform indexes gaps; new joiners can query for them; the team can adjudicate which ones to close.

3 · Pin a "recent revisions" view in every agent UI.​

The hardest thing for a new joiner to discover is what has changed recently. Expose the revision window prominently; it is the highest-signal slice of the codex on day one.

4 · Use scope freezes when archiving, not deletions.​

When a project ends, freeze its scope rather than deleting it. Frozen scopes remain queryable; a future incident or audit can still walk their codex.