Skip to content

Example: separate harness repo, partially bound

Synthetic example — constructed to illustrate the report shape.

These numbers are invented to show what the report looks like when a shared harness binds in some subjects and not others. They are not a reading of any real estate. The self-assessment is the real one.

A platform team owns platform-harness. Four services are meant to be governed by it. Two are, one is pinned a year back, and one is not bound at all.

This is the shape that matters most in practice, because on an architecture slide all four services look identical.

The manifest

version: 1
team: payments-tribe

habitats:
  - id: platform-harness
    kind: repo
    path: ./platform-harness
    provides: [HARNESS.md, AGENTS.md, ci, hooks]

subjects:
  - id: orders-api
    path: ./orders-api
    habitat: platform-harness
  - id: billing
    path: ./billing
    habitat: platform-harness
  - id: reporting
    path: ./reporting
    habitat: platform-harness
  - id: legacy-batch
    path: ./legacy-batch
    habitat: platform-harness

Coverage ledger

Subject Status Report date Age Note
orders-api assessed 2026-08-01 13d
billing assessed 2026-07-29 16d
reporting assessed 2026-08-06 8d harness pinned 412d
legacy-batch assessed 2026-08-02 12d harness declared, not bound

4 of 4 subjects assessed; 0 stale; 0 unreachable.

Dimension matrix

inherited · inherited-unbound · unmarked is local · * weakest in column

Subject AB AI WF OM TP OR OA HR WP AC AD Te Ob Gv
orders-api 3 3 4ᶦ 3 4ᶦ 3 3 3 3 3 3 4ᶦ 3 4ᶦ
billing 3 3 4ᶦ 3 4ᶦ 3 3 3 3 2 3 4ᶦ 2 4ᶦ
reporting 3 2 3ᶦ 3 3ᶦ 3 3 3 2 2 3 3ᶦ 2 3ᶦ
legacy-batch 2* 1* 1ᵘ* 2* 1ᵘ* 2* 2* 2* 2* 1* 2* 1ᵘ* 1* 1ᵘ*

AB Agent behaviour · AI Agent input · WF Workflow · OM Operating model · TP Teams provide · OR Output role · OA Output artefact · HR Humans review · WP Work patterns · AC Agent composition · AD Agents… · Te Testing · Ob Observability · Gv Governance

The marks are the finding. legacy-batch declares platform-harness in the manifest, but nothing in the repository reaches it: its CI defines its own workflows rather than reusing the shared ones, its AGENTS.md restates rules rather than deferring, and there is no pin. Four dimensions the harness would have supplied are capped at what local evidence supports.

reporting is bound — via a submodule pin — but the pin is 412 days old. It is assessed against the pinned revision, which is why its inherited dimensions sit a level below the two current subjects.

Gap table

Subject Habitat maturity mean Cognitive level Gap Regime
orders-api 3.29 3 −0.29 Coherent
billing 3.14 3 −0.14 Coherent
reporting 2.71 3 +0.29 Coherent
legacy-batch 1.50 3 +1.50 Ambition outpaces enablement

Spread

−0.29 (orders-api) to +1.50 (legacy-batch) — range 1.79.

One team, one way of working, four habitats — and the range is wide. The same engineers who sit comfortably Coherent in orders-api are working a full level and a half ahead of their environment in legacy-batch. That is not a capability difference. It is an environment difference, and it is the reason a single-repository assessment of orders-api would have reported a healthy team and missed this entirely.

Split ceiling

Common weak — Observability, Agent composition. Weak in three of four subjects, including subjects where the harness does bind. The shared harness does not supply these, so binding it harder will not fix them. Enablement backlog.

Locally weak — legacy-batch: Workflow, Teams provide, Testing, Governance. These read as local weaknesses but they are not: all four are inherited-unbound. The capability exists in platform-harness already. This is a binding failure, not a team failure, and it belongs to whoever owns the harness.

Confidence

56 placements: 32 observed, 20 inferred, 4 asked. The four asked placements were spent on legacy-batch, confirming that the harness is genuinely not wired up rather than bound by a mechanism the checklist does not recognise — the report does not assert inherited-unbound without that confirmation.

What this estate should do next

Bind the harness in legacy-batch. This is the cheapest uplift available anywhere in the estate, and it is not a build. Nobody needs to write a testing policy or a governance model — platform-harness has both. Somebody needs to make legacy-batch reuse the shared workflows and defer to the shared conventions. Four dimensions move on the day it lands.

Refresh the reporting pin. It is bound, which is the hard part. It is 412 days behind, which is the easy part.

Then lift observability and agent composition in the harness itself. Only after the binding work — doing it first would improve a harness that half the estate cannot see.

Note the ordering. The instinct with a matrix like this is to start with the lowest row and treat legacy-batch as a struggling team. The provenance marks say otherwise: the estate's biggest number is a wiring problem, and treating it as a capability problem would have that team rebuild what already exists twenty metres away.


Portfolio regime naming — Federated, Distributed, Fragmented, Islanded — arrives in a later release. This example shows the shape a distributed estate produces, without asserting the label.

View this report as a shareable HTML page — a single self-contained file, no network assets, works opened straight from disk.