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 — architecture & how-it-works docs, DB/platform tooling. Part of the hatch → code docs migration tracking issue. Full plan: ~/workspace/scratch/docs-migration-plan-2026-11-04/plan.md (§4.3).
Rule: copy, never delete. Sources stay as historical copies; one-line Canonical: headers on living docs after landing.
Files (25)
From hatch-personal (= dezren39/hatch) → code:
tools/supabase-exec.py → pkgs/fleet/ or pkgs/database/ — fleet-wide Supabase SQL helper. Load-bearing dependency: the sandbox quota-recorder (which STAYS in hatch by Drew's ruling) calls this script at its hatch path. The copy in code becomes canonical for new work; the hatch copy must remain functional until quota-recorder is repointed — do that repoint in this batch or explicitly defer it in the PR.
tools/cloudflare/wrangler-authed.py, tools/cloudflare/wrangler-secret.py → pkgs/cf/ — Cloudflare tooling (resolves creds via the sandbox bws route; the route doc stays in TOOLS.d).
library/toolchains/ (13 files excl. binary-backup.md, which is in the backups batch): bun, chromium-playwright, cloudflare-wrangler, git, go, index, muse-cli, nix, onepassword-op, rust, tailscale, uv, bitwarden-bws → doc/platform/toolchains/. Borderline pair: nix.md (overlaps TOOLS.d/nix.md) and muse-cli.md — flagged for Drew; move with the cluster unless he says otherwise.
notes/research/muse/ (6 files: README, mission, report, notes/source-01, notes/source-02, social-posts) → doc/platform/ — research into how the Muse platform works (borderline: platform research vs sandbox doc; Drew's "how Muse tasks work" example tips it to MOVE)
Atlas note
There are no Atlas docs in the repos — Atlas material is swept by the separate Atlas batch issue. The architecture basics Drew named (Supabase, Cloudflare, fleet layout) land via this batch + the fleet batch.
Expectation: docs migration batch (25 files); triage done = batch scoped, load-bearing dependency and flagged items recorded below
Links: plan ~/workspace/scratch/docs-migration-plan-2026-11-04/plan.md §4.3 (verified present on the sandbox today, with the quota-recorder dependency noted in the plan)
Triage findings (verified fresh 2026-11-05 via gh + local plan check):
Batch scope (25 files, copy — never delete):tools/supabase-exec.py and tools/turso-exec.py → pkgs/fleet/ or pkgs/database/; tools/cloudflare/wrangler-authed.py + wrangler-secret.py → pkgs/cf/; library/toolchains/ (13 files, excl. binary-backup.md) → doc/platform/toolchains/; library/plans/ai-plans.md + hardware.md and library/inventory-summary.md → doc/platform/; notes/research/muse/ (6 files) → doc/platform/. Sources stay as historical copies; living docs get a one-line Canonical: header after landing.
Load-bearing dependency (the one real risk in this batch): the sandbox quota-recorder STAYS in hatch (Drew's ruling, plan §4.3/line 268) and calls tools/supabase-exec.py at its hatch path. The code copy becomes canonical for new work, but the hatch copy must remain functional until the quota-recorder is repointed — the implementing PR must either do that repoint or explicitly defer it in the PR body. Breaking the hatch copy breaks quota recording.
Flagged/borderline items, with the issue's own defaults (no Drew decision needed to proceed):library/toolchains/nix.md (overlaps TOOLS.d/nix.md) and muse-cli.md are flagged for Drew but the instruction is to move them with the cluster unless he says otherwise; library/inventory-summary.md is borderline → doc/platform/; the notes/research/muse/ set is tipped to MOVE by Drew's "how Muse tasks work" example. The implementing PR should name these four calls explicitly so the flag is visible at review.
Destination decision still open inside the batch:pkgs/fleet/ vs pkgs/database/ for the two SQL helpers — pick one in the PR and state why; everything else has a named destination.
Atlas: explicitly out of scope — no Atlas docs are in the repos; Atlas material is swept by the separate Atlas batch issue.
Ordering / blockers: none external. This batch is independent copy work; it does not gate, and is not gated by, the other migration batches beyond the shared copy-never-delete rule.
Triage complete for task 10349; issue #5980 stays open for the migration PR.
Activity
developing-today commentedon Nov 5, 2026
CODE-1646
dezren39 commentedon Nov 5, 2026
🤖 Agent note · done (triage)
~/workspace/scratch/docs-migration-plan-2026-11-04/plan.md§4.3 (verified present on the sandbox today, with the quota-recorder dependency noted in the plan)Triage findings (verified fresh 2026-11-05 via
gh+ local plan check):documentation,task/triage,queue/queued. No hard-stop labels (held,user/owned,do-not-merge,do-not-auto-edit) and nouser/review/user/escalated. Only comment is an automated Kaneo cross-link.tools/supabase-exec.pyandtools/turso-exec.py→pkgs/fleet/orpkgs/database/;tools/cloudflare/wrangler-authed.py+wrangler-secret.py→pkgs/cf/;library/toolchains/(13 files, excl. binary-backup.md) →doc/platform/toolchains/;library/plans/ai-plans.md+hardware.mdandlibrary/inventory-summary.md→doc/platform/;notes/research/muse/(6 files) →doc/platform/. Sources stay as historical copies; living docs get a one-lineCanonical:header after landing.tools/supabase-exec.pyat its hatch path. The code copy becomes canonical for new work, but the hatch copy must remain functional until the quota-recorder is repointed — the implementing PR must either do that repoint or explicitly defer it in the PR body. Breaking the hatch copy breaks quota recording.library/toolchains/nix.md(overlaps TOOLS.d/nix.md) andmuse-cli.mdare flagged for Drew but the instruction is to move them with the cluster unless he says otherwise;library/inventory-summary.mdis borderline →doc/platform/; thenotes/research/muse/set is tipped to MOVE by Drew's "how Muse tasks work" example. The implementing PR should name these four calls explicitly so the flag is visible at review.pkgs/fleet/vspkgs/database/for the two SQL helpers — pick one in the PR and state why; everything else has a named destination.Triage complete for task 10349; issue #5980 stays open for the migration PR.