[gregoryfoster]

Closes #23.

Docs only: no restart.

• Spec §6's table gains a Status column. Each row is marked done, resolved upstream, or moot, with a link. The planned Change text is kept as approved, and corrections to it go in Status. Review and evidence: Spec §6: the cross-repo notices never posted (replicator charter, archiver#179, cannobserv, observo token) #23 (comment)
• Archiver: comment posted with the operator's approval: Processing infrastructure for canonical derived artifacts — contract-first, leaning Observo (successor to replicator#69) archiver#179 (comment)
• Replicator: moot, no post. The committed charter names no processor home, and the row's quoted "Observo transforms" is in no Replicator file.
• Cannobserv: resolved upstream by cannobserv#503, which is on main and unreleased (the pinned 0.19.7 still carries the old docstrings). The row named the wrong file: the stale text was co-core's streams.py and test_streams.py, not CHANGE_BUS.md.
• Observo, broker, watcher, GCP / tailnet: done.
• Spec header: the contract pointer notes that by v0.19.7, the pinned version, the section lives in docs/change_bus/process.md.

Review (CR --fix, two rounds), one commit each:

• CR 1 restores the §6 heading and the sentence that the first commit's splice overwrote.
• CR 2–6 correct the cannobserv location, the header version, the kept Change text, the unreleased note, and the tailnet evidence.
• CR 8–9 are polish. CR 7 is this description.
• ea328ee's message says the stale text was in change_bus/process.md. CR 2 supersedes that.

Gate: ruff check, ruff format --check, and pytest -m 'not integration' pass locally (219). The earlier CI run was cancelled by a GitHub Actions incident: no runner was assigned, 0 steps ran, and the 15-minute timeout hit.

🤖 Generated with Claude Code