GloriousFlywheel Active Owned Repo Adoption Board 2026-04-17
Snapshot date: 2026-04-17
Purpose
Turn the broad adoption plan into a concrete active-repo board.
This note is the Phase 0 execution board for the current owned-repo surface:
- which active repos already show GloriousFlywheel runner or cache adoption
- which repos are the real dogfood blockers
- which repos are merely adjacent and should not steal focus
Companion notes:
- gloriousflywheel-adoption-homogeneity-and-dogfood-plan-2026-04-17.md
- gloriousflywheel-wave0-wave1-classification-matrix-2026-04-17.md
- gloriousflywheel-downstream-adoption-and-migration-kit-2026-04-16.md
- gloriousflywheel-first-tranche-repo-contracts-2026-04-16.md
- gloriousflywheel-massageithaca-dogfood-playbook-2026-04-17.md
Evidence Basis
This board is grounded in:
- current visible
gh repo listactivity for non-fork repos intinyland-incandJesssullivan - live workflow adoption evidence gathered on 2026-04-17
- existing dogfood, benchmark, and downstream-adoption notes already in this repo
Important limitation:
- GitHub code-search rate limits hit during the broader cache census
- runner-label evidence is stronger than cache-specific evidence for the full active set
- where cache or FlakeHub posture is not yet directly proven, the board says so explicitly instead of inferring too much
Triage Classes
dogfood-blocker
Use this when the repo directly affects whether GloriousFlywheel can make an honest product claim right now.
Typical reasons:
- source-of-truth repo drift
- named product dogfood lane
- benchmark canary
- platform bug surfaced by a real consumer
active-consumer
Use this when the repo is a real runner or cache consumer but not the next place where platform truth is blocked.
adjacent-surface
Use this when the repo shapes templates, publication, or platform adjacency without being the highest-value runner-contract proof point itself.
hold
Use this when the repo is active but there is not yet enough evidence that it should be in the near-term GloriousFlywheel migration queue.
Active Board
| Repo | Active role | Current GF evidence | Triage class | Near-term action |
|---|---|---|---|---|
tinyland-inc/GloriousFlywheel |
source-of-truth platform repo | mixed hosted and self-hosted CI on default branch; strongest ARC test flow still branch-gated; live self-dogfood exists but is not the routine proof path | dogfood-blocker |
move normal runner-contract proof onto default-branch CI and stop letting source-of-truth drift hide behind hosted checks |
Jesssullivan/MassageIthaca |
named product dogfood repo | repo-owned PRIMARY_LINUX_RUNNER_LABELS_JSON lane for core CI; beta and alpha post-deploy jobs were the hosted escape hatch |
dogfood-blocker |
land and verify the self-hosted beta-validation PR path; use this repo as the proof that product repos do not need hosted escape hatches |
tinyland-inc/lab |
cache and builder canary | explicit tinyland-nix runner use, setup-flywheel in Bazel builds, and live FlakeHub cache and publish actions in the Nix workflow |
dogfood-blocker |
treat lab as the first benchmark and cache-contract canary, and define one canonical path instead of treating every lane as equally authoritative |
tinyland-inc/tinyland.dev |
representative org product repo | mixed truth: Bazel CI and staging deploy are still on ubuntu-latest, while the heavy Nix build path runs on tinyland-nix and pushes to Attic |
dogfood-blocker |
decide whether this repo is supposed to prove GloriousFlywheel runners or just GloriousFlywheel caches, then align the authoritative CI path with that answer |
Jesssullivan/XoxdWM |
strongest user-owned public canary | explicit tinyland-* labels and visible setup-flywheel usage; still mixes stable shared lanes with hardware and compatibility flows |
dogfood-blocker |
finish normalizing the shared fast path and keep hardware-only jobs clearly outside the public contract |
tinyland-inc/blahaj |
infra-adjacent operational dependency | real shared-runner usage exists on tinyland-docker and tinyland-dind, but large parts of the repo are cluster-ops or hosted utility flows rather than a clean downstream contract |
dogfood-blocker |
keep clarifying cluster-ops ownership versus stable shared-runner consumption and stop treating the repo as one homogeneous dogfood surface |
Jesssullivan/acuity-middleware |
active product-adjacent consumer that surfaced platform defects | package CI and publish already use a verified repo-specific self-hosted lane through the shared template, while Modal deploy remains explicitly hosted | dogfood-blocker |
treat this repo as proof for workspace and checkout hygiene work in ci-templates and the runner envelope, then decide whether hosted deploy is a durable exception |
Jesssullivan/yt-text |
heavy Nix dogfood canary | core CI runs on shared self-hosted Nix labels, but docs deploy stays GitHub-hosted and cache/bootstrap assumptions are still mostly implicit in runner state | dogfood-blocker |
use it to prove heavy-Nix lane value under measured load and make the lane contract explicit in workflow code |
tinyland-inc/tinyvectors |
active org consumer | shared JS and Bazel package template is in use, but workflow code does not pass runner labels, so current behavior is hosted unless repo variables override it | active-consumer |
verify actual repo-variable use before counting this repo toward runner adoption and otherwise treat it as a hosted template consumer |
tinyland-inc/betterkvm |
active org consumer | current default-branch CI is still fully ubuntu-latest; it consumes Nix and Attic today, but not GloriousFlywheel runners yet |
active-consumer |
classify it as a hosted cache consumer for now and schedule explicit tinyland-nix adoption before counting it toward runner penetration |
Jesssullivan/scheduling-kit |
active platform-adjacent product repo | CI and publish use a verified repo-specific self-hosted lane through the shared JS and Bazel template, but that template is still cleanup-based and hosted-fallback capable | active-consumer |
use it as the first clean pilot for js-bazel-package V2 and promote it from repo-variable indirection into an explicit documented dedicated-lane contract |
Jesssullivan/tinyclaw |
active user consumer | inspected build, PR, GHCR, release, and Nix flows are all GitHub-hosted today; it is active, but not a GloriousFlywheel runner consumer on main |
active-consumer |
keep it out of runner-adoption claims until a real self-hosted or cache-contract lane lands |
Jesssullivan/remote-juggler |
active user consumer | strong Nix and Attic usage is real, but the inspected default-branch CI and release paths are still GitHub-hosted rather than GloriousFlywheel-hosted | active-consumer |
separate cache-consumer status from runner-consumer status and only promote it once a real default-branch self-hosted lane exists |
Jesssullivan/tummycrypt |
cache-adjacent active repo | strong hosted Nix and Attic usage is real, but the inspected default-branch surfaces are GitHub-hosted rather than GloriousFlywheel-hosted | active-consumer |
keep it in the cache-contract wave and only promote it into runner-adoption metrics after self-hosted default-branch use is real |
tinyland-inc/ci-templates |
reusable CI template surface | recent active repo; adjacency is strong because it can encode or spread runner and cache patterns | adjacent-surface |
align templates with GloriousFlywheel contract after the contract itself is stable enough to template |
tinyland-inc/.github |
org-level policy and default automation surface | active org-level repo, but not a direct runner proof point in current evidence | adjacent-surface |
use it later for policy rollout after runner-contract language is settled |
tinyland-inc/bazel-registry |
Bazel ecosystem adjacency | active and strategically relevant, but not yet a named runner-contract or dogfood repo in current evidence | adjacent-surface |
keep it out of the first migration wave and revisit once Bazel dogfood is benchmarked more clearly |
Jesssullivan/jesssullivan.github.io |
personal lane consumer, not repo-owned baseline | current notes explicitly say personal-nix and personal-docker are not repo-owned baseline lanes |
adjacent-surface |
keep separate from the baseline product claim and avoid letting personal-lane policy pollute the public contract |
tinyland-inc/FinanceBro |
active internal repo | currently active by push date, but no strong runner or cache evidence in the current GloriousFlywheel notes | hold |
verify workflow shape before pulling it into the migration queue |
Jesssullivan/tinyland-reorg |
active private repo | active by push date, but no current runner-contract evidence in the active notes | hold |
leave out of the near-term queue until there is explicit platform relevance |
Current Waves
Wave 0: Dogfood Blockers
These repos should drive the next execution window:
GloriousFlywheelMassageIthacalabtinyland.devXoxdWMblahajacuity-middlewareyt-text
Reason:
- together they cover source-of-truth drift, product dogfood, user canaries, builder/cache proof, infra-adjacent consumption, workspace hygiene, and heavy-Nix scaling truth
Wave 1: Active Consumers
Move here after Wave 0 is cleaner:
tinyvectorsbetterkvmscheduling-kittinyclawremote-jugglertummycrypt
Reason:
- these repos appear active and relevant, but they do not currently define the credibility of the platform claim
Wave 2: Adjacent Surfaces
Do not let these repos lead the migration narrative:
ci-templates.githubbazel-registryjesssullivan.github.io
Reason:
- they shape policy, packaging, or future builder/publication surfaces
- they should consume the contract after the contract is more stable
Immediate Outputs Needed
1. Repo Classification Matrix
For Wave 0 and Wave 1, add one row per repo covering:
- runner pattern
- workspace pattern
- Nix bootstrap mode
- Attic expectations
- Bazel expectations
- hosted escape hatches
- rollback path
The first working version now lives in:
2. Benchmark Pack Mapping
The first measured pack should use:
GloriousFlywheellabtinyland.devXoxdWMyt-textMassageIthacafor product-repo post-deploy truth
3. Exception Register
Track every current exception explicitly:
- hosted jobs on dogfood repos
- raw
[self-hosted, ...]selectors outside documented hardware cases - personal lanes mistaken for baseline product lanes
Success Criteria
- the near-term queue is no longer “all active repos”
- Wave 0 repos each have one named blocker or proof obligation
- the platform can explain why a repo is in
dogfood-blocker,active-consumer,adjacent-surface, orhold - the next migration wave can be driven by evidence rather than by whichever repo was edited most recently
Recommendation
The next work should stay concentrated on Wave 0 until all of these are true:
GloriousFlywheeldefault-branch CI contains routine self-hosted proofMassageIthacano longer depends on hosted beta validationlabandyt-textprovide real cache and heavy-Nix benchmark evidenceXoxdWMis a clean public user-repo contract exampleacuity-middlewarecheckout hygiene work is closed at the platform layer
Only then should Wave 1 become the default execution lane.