Skip to content

specanchor RFC: proposed → accepted (status truth after #165) - #166

Merged
omercelikdev merged 1 commit into
mainfrom
specanchor-rfc-accepted
Aug 16, 2026
Merged

omercelikdev merged 1 commit into
mainfrom
specanchor-rfc-accepted

Conversation

@omercelikdev

Copy link
Copy Markdown
Collaborator

Two-line ledger truth: the owner accepted the composition RFC by merging #165, so the RFC header and the index row now say so, and DoD item 1 is checked. ledger-check.sh + docs-freshness.sh green.

🤖 Generated with Claude Code

Owner accepted 2026-08-16 (#165 merged). Remaining DoD: the ledger records
(shipped in #165) and the rehearsal (open-threads T20).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@omercelikdev
omercelikdev merged commit 5008ba6 into main Aug 16, 2026
@omercelikdev
omercelikdev deleted the specanchor-rfc-accepted branch August 16, 2026 20:35
omercelikdev added a commit that referenced this pull request Sep 8, 2026
…ft a gap (#253)

Every repository named in the delivery-cycle RFC now carries the layer:
mockifyr #417, qorpe/ui #49, mediant #166, api-portal #74, factoring-poc #7,
specdrift #13, specanchor #1. Only coexist's regeneration and D6's spec
chain are left.

The two repositories that carried NO agent layer also carried no docs gate,
and writing one for each found live drift in both — recorded in the row
because a gate that never caught anything is a gate nobody trusts:

  specdrift   README pinned 0.4.1 while the csproj shipped 0.4.3, so the
              Docker tag and Action ref in the run-it-anywhere table were
              two releases stale. `--help` printed `specdrift mcp` below
              the prose about the other two verbs.
  specanchor  the ledger called the shipped licence gate "still owed"; the
              README's "Next:" named the gates, the CLI and MCP, all three
              built and exercised in CI; `specanchor mcp` was documented
              nowhere at all; one section was written in Turkish, in a
              repository whose whole product is an auditable trail read by
              client-side reviewers.

T28 records what the sweep exposed. cycle.md now lives in nine places and
nothing compares the copies ACROSS repositories — skills-parity.sh only
compares the three inside this one. The cycle has already been revised once
(#252 made §6 stack-neutral); the next revision reaches adopters only if a
person remembers, and `goldpath init` does not ship the skills family
because template-completion D5 assigns that to the unbuilt transformation
pack. The thread carries the owner's no-extra-spend constraint in its proof
column, because that is what rules out the obvious answers.

All five gates green.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant