Wave 0: regenerate canonical flake×Guix×container inventory worklist

[hyperpolymath]

Parent: #102. Refs #101.

Produce the durable worklist the campaign slices waves from. The triage inventory existed only in a sub-agent transcript and was never committed.

Deliverable

A committed artifact (suggest rhodium-standard-repositories/spec/nix-retirement-inventory.adoc) with, per estate repo:

repo | flake.nix path(s) | guix.scm/manifest.scm path(s) | .guix-channel? | sealed-container? (Containerfile count) | provisional verdict

Provisional verdict ∈ { pure-mirror→remove (pending per-repo dep diff), load-bearing→keep, guix-channel-only→keep+flag, monorepo→handle-at-source, out-of-scope (no Guix), excluded }.

Method (supervised, foreground — bg agents are sandbox-blocked for this)

1. gh repo list hyperpolymath --limit 1000 → enumerate.
2. Per repo probe flake.nix, guix.scm/manifest.scm, .guix-channel, Containerfile presence.
3. Apply gate rules 1–6 from Campaign: retire Nix-mirror across consumer repos (Guix-primary ruling) #102 for the provisional verdict (NOT the final delete decision — that needs per-repo flake-input vs Guix vs container diffing at wave time).
4. Exclude standards, echidna.
5. Commit as a PR (Refs #101), squash+auto.

Then

Slice waves 1..N (~20 single-package repos each) as further child issues of #102.

[hyperpolymath]

Wave-0 artifact is committed; ready to close

The deliverable named in this issue exists in main at rhodium-standard-repositories/spec/nix-retirement-inventory.adoc (commit 5f2f8b2, 462 lines). It carries the per-repo repo | flake.lock | guix-paths | container-count | provisional-verdict table across 379 estate repos: 277 candidates sliced into 14 waves (#106–#119), 8 monorepos (handle-at-source), 2 excluded (standards, echidna), 92 out-of-scope.

The wave-time diff has now been done (Refs #102)

The artifact explicitly said the final per-repo flake-vs-(Guix ∪ container) diff happens at wave time. That work has now been completed across Waves 1–14 (~280 in-scope repos analysed, verdicts posted to each wave issue):

outcome	count	meaning
pure-mirror → REMOVE (verified first-hand)	1	hypatia only — PR open at hyperpolymath/hypatia#289
load-bearing → KEEP+DEP	~6	007, boj-server, cyo, echidnabot, file-soup, game-server-admin, ubicity
scaffold-stub guix.scm → KEEP+FLAG (out of scope, scaffold-leak debt)	~270	dominant pattern: empty (source #f) / TODO-only (inputs (list)) / unrendered {{PROJECT_NAME}} templates
over-called REMOVE → demoted to KEEP+DEP on first-hand verification	2	proof-of-work (flake-only just, cargo-audit, cargo-deny, mold, etc.); tma-mark2 (flake-only pkg-config, nickel, just)

Campaign-level finding

The Nix-mirror campaign's effective surface is far smaller than the candidate count suggests: ~270 of the 277 candidates are scaffold-stub debt (empty guix.scm paired with template flake.nix), not real mirrors — they fail the "REAL inputs" half of the gate. That belongs on a scaffold-leak remediation track, not the Nix-mirror retirement track. The genuine pure-mirror harvest across the whole estate is 1.

Recommend closing this Wave-0 issue (the inventory exists + has been operated on; further work is per-wave, tracked under #106–#119 and the new scaffold-debt track).

Refs #102 #101.

[hyperpolymath]

Closing as completed: Nix-mirror campaign closure documented in #149 (merged). See campaign closure doc and hypatia#289 (merged) for the consolidated record. All 12 retire/annotate PRs merged. Reopen if a specific wave deliverable is missing.