You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Docs migration batch — darwin-nix non-flake docs → code. Part of the hatch → code docs migration tracking issue. Full plan: ~/workspace/scratch/docs-migration-plan-2026-11-04/plan.md (§5). Repo rule (Drew): darwin-nix docs should be about the flake only; documentation about anything else gets its canonical home in code. Copy, never delete — darwin-nix keeps historical copies; add one-line Canonical: headers after landing. Functional flake code/config that merely uses a topic stays (see findings).
Files (32)
planning/plans/mcpx-extraction.md, planning/plans/mcpx-identity-transport.md, planning/plans/mcpx-oauth-authorization-server.md → doc/fleet/ (mcpx designs; mcpx canonical home is pkgs/mcpx)
docs/mcp.md → doc/fleet/ (mcpx + MCP server inventory doc)
mcpx — docs above move. Functional stays: mcpx.config.nix (borderline: config the flake consumes; Drew said "anything about mcpx" — flagged in the tracking issue for his call), pkgs/npm-mcp-servers/, mcp-clients.py + activation scripts/tests, openchamber.nix references. Mcpx sections inside staying docs (ROADMAP.md — its largest area — ARCHITECTURE.md, README.md) should be trimmed to pointers at the code copies when this batch lands.
ID project — no ID documentation in darwin-nix. (mcpx-identity-transport.md is mcpx, counted above.)
Atlas — none. All "atlas" hits are Atlassian (pkgs/twg-cli, atlassian MCP entries).
Cloudflare — zero hits in the whole tree (grep-verified).
Turso — zero hits in the whole tree (grep-verified).
Done looks like
Copies committed in code via PR, pointer headers in darwin-nix copies, ROADMAP/ARCHITECTURE mcpx sections trimmed to pointers, issue closed with PR link.
Drew's rulings on the mcpx borderline calls — 2026-11-05 00:05 CST review of the migration plan
History note from Drew: mcpx was born in darwin-nix and later moved elsewhere; not all doc locations were updated when it moved. That is the origin of the mcpx contamination in this repo — stale locations, not deliberate placement.
Rulings for this batch:
(a) mcpx.config.nix STAYS as functional config. Agreed with the plan's borderline call.
(b) ROADMAP.md / ARCHITECTURE.md: anything in them that is NOT about nix-darwin gets MOVED out. This is stronger than the plan's "trim to pointers after the batch" — Drew's word is moved, and this ruling supersedes the plan's borderline call for these two files. (The sections' subject matter lands wherever this migration puts mcpx / agent-workflow docs; nothing non-nix-darwin stays behind as prose.)
(c) archive/ (207 vendored third-party skill files): NO conclusion yet — explicitly deferred. Drew wants a dedicated discussion about the archive before anything is decided; do not move, delete, or reclassify it as part of this batch. It has been added to Drew's pending-questions queue (item darwin-archive-207) so the discussion is owed, not forgotten.
Activity
developing-today commentedon Nov 5, 2026
CODE-1648
dezren39 commentedon Nov 5, 2026
Drew's rulings on the mcpx borderline calls — 2026-11-05 00:05 CST review of the migration plan
History note from Drew: mcpx was born in darwin-nix and later moved elsewhere; not all doc locations were updated when it moved. That is the origin of the mcpx contamination in this repo — stale locations, not deliberate placement.
Rulings for this batch:
mcpx.config.nixSTAYS as functional config. Agreed with the plan's borderline call.ROADMAP.md/ARCHITECTURE.md: anything in them that is NOT about nix-darwin gets MOVED out. This is stronger than the plan's "trim to pointers after the batch" — Drew's word is moved, and this ruling supersedes the plan's borderline call for these two files. (The sections' subject matter lands wherever this migration puts mcpx / agent-workflow docs; nothing non-nix-darwin stays behind as prose.)archive/(207 vendored third-party skill files): NO conclusion yet — explicitly deferred. Drew wants a dedicated discussion about the archive before anything is decided; do not move, delete, or reclassify it as part of this batch. It has been added to Drew's pending-questions queue (itemdarwin-archive-207) so the discussion is owed, not forgotten.