Skip to content
Octopus Octopus Octopus

audit-fleet

audit-fleet is the cross-repo view a manager over 6+ repos otherwise lacks. octopus setup is per-repo and workspace: shares rules, but nothing shows which repos lag the standard. This audit does: it resolves the fleet and reports, per repo, how the actual Octopus config compares to the target the team declared in fleet.yml — baseline + stack profile + adoption tier — plus Octopus version and CONTEXT.md/ADR adoption.

It is the detect half of the fleet pair: audit-fleet finds the drift, fleet-bootstrap closes it. The report is built to feed fleet-bootstrap --from-audit. The audit is signal-only and read-only — it never mutates a repo.

For the full fleet model, see Fleet setup flow.

Drift is measured against the declared target

Because the fleet is multi-stack with varying adoption, “drift” only means something relative to what a repo is supposed to be. audit-fleet computes each repo’s target the same way fleet-bootstrap would — baseline ∪ profile ∪ tier — and compares:

  • Adoption tier — effective tier (hooks/precommit/qualityWorkflow) vs declared tier (below target = drift; above is flagged too).
  • Stack profile / bundles — present vs the profile’s expected bundles; detected stack vs declared profile.
  • Octopus version — installed vs latest.
  • Rules drift — *.local.md + workspace rules vs the baseline.
  • Encode adoption — CONTEXT.md present? ADR count?

It reuses the single-repo checks from audit-config rather than reimplementing them — it is the cross-repo analog.

The report

A per-repo table (version · target tier → actual · profile match · missing bundles · CONTEXT.md? · ADRs · drift flags) plus a drift-hotspots rollup — the widest gaps first, so the manager sees the biggest standardization gaps before the detail:

Hotspots: 2/4 behind version · 3/4 no CONTEXT.md · 1/4 below target tier
→ remediate with `fleet-bootstrap --apply` (or `--from-audit` this report)

Execution model

v1 runs locally against checked-out repos (no infra), consistent with fleet-bootstrap. An org-level GitHub Action scanning remotes is a deferred follow-up in the fleet execution model.

Composes with

  • fleet-bootstrap — the remediation half; shares the fleet list; consumes this report via --from-audit.
  • audit-config — the single-repo checks this audit reuses fleet-wide.
  • audit-grounding — depends on the CONTEXT.md/ADR adoption this surfaces.

Source: skills/audit-fleet/SKILL.md