[EHotwagner]

Records the published FS.GG.Governance.Cli / FS.GG.Governance.FSharpSurfaceCommand 1.15.0 in the registry. This is the registry-reconcile part of condition 4 in the GO comment on FS-GG/FS.GG.Governance#423 (comment 6361365085).

The changes are text only. No contract version, coherence flag or package pin changes:

• registry/dependencies.yml, governance-cli-handoff-consumer-published: the summary now says the feed's newest version is 1.15.0 and describes what 1.15.0 changes. resolved_by and impact gain the 1.15.0 tag, PR, publish run and readback evidence.
• registry/CHANGELOG.md: a new dated entry.
• docs/registry/compatibility.md: a short runtime-release paragraph after the 1.11.0 one. check-projection.py passes and the projections are unchanged.

Evidence

• Publication: annotated tag v1.15.0 points to 282741b1 (Governance PR chore(deps): update actions/checkout action to v7 #472). Publish run 39928935417 succeeded.

• Readback: run 39935018723 is read-only and was merged in Governance PR [cross-repo] Implement ADR-0032: exclude the SDK's library-packs so FSharp.Core resolves from ONE source, then re-pin every lock as a coherent set #473 after independent review.

  • Cli: org 0b0ebefd… and nuget.org 9eb3eed6… match on all 257 entries.
  • FSharpSurfaceCommand: org 7cecc0a8… and nuget.org 4c9147d6… match on all 38 entries. The org hash equals the run's recorded producer hash.
  • nuget.org adds only .signature.p7s, and both nuspecs name commit 282741b1.

• Installed journey: I cold-installed both tools from the nuget.org bytes into isolated tool paths. The installed CLI then passed the provider-mode SDD 2.4.0 journey in the fdev container on Home (glibc 2.44):

  • positive case: exit 0;
  • missing, stale, mismatched and exit-2 evidence: exit 1;
  • preserved refusals: exit 3.

  A separate run as eugen on Home will be admitted on its own.

Merge hold: released by the coordinator.

Review: an independent fsgg-critic-best review asked for changes on 7e9928f5 (wording: attribution of the journey, the profile name, which packages were installed). It accepted e6a7eb3f.

architecture-map: unaffected

🤖 Generated with Claude Code

[EHotwagner]

fdev: merge note for head 47f93ed0.

Rebase.

• I rebased twice onto a busy main (3d45db23, then b4933bfa).
• git range-diff shows both commits patch-identical (=) to the reviewed 7e9928f5/e6a7eb3f.
• b4933bfa (Prepare independent Host 0.9 to admit Claude transcript usage (Refs #4640) #4834, Host 0.9) touched other rows of registry/CHANGELOG.md and dependencies.yml; the rebase merged cleanly without conflicts.
• check-projection.py passes, and generate-projections leaves the tree clean.
• I am treating this as non-material for the independent ACCEPT on e6a7eb3f.

Checks.

• All required checks pass, and so does routine-eligibility.
• The only red check is the non-required feed. It fails because new-sdd-workspace 0.22.0 has been published while the registry still declares 0.21.0. That belongs to the Creator/Wizard 0.22 reconcile lane (Reconcile accepted Creator 0.22 publication in the registry and Atlas (Refs #4811) #4838) and is unrelated to this PR.