Skip to content

ecosystem-adoption.md

Latest commit

 

History

History
365 lines (346 loc) · 27 KB

ecosystem-adoption.md

File metadata and controls

365 lines (346 loc) · 27 KB
Raw

Ecosystem Adoption and Derivatives

简体中文

This maintainer-observed inventory links public evidence of projects using, integrating, studying, or reimplementing LoopX. Start with workflows and integrations for concrete use; proposals and deferred adoption records work that has not become an adopted runtime.

Evidence boundary: inclusion is a factual record, not an endorsement or a production-deployment claim. A merged PR proves that code or documentation landed; a published release, bounded pilot, and open proposal prove different things. Validation results in linked reports are their authors' reports, not independent reproductions by this inventory.

ADOPTERS.md is the separate, voluntary directory where projects and users describe their own use. Its empty registration table does not mean there is no observed use: the evidence below does not depend on an owner submitting a directory entry.

1. Workflows and Integrations

  • CGC 2046 (CodingGirlsClub) — merged PR #838 records a move to LoopX with Codex CLI for development orchestration; merged PR #841 adds conditional merging to develop while retaining human control for specified sensitive changes, main, and releases. Status: development-workflow configuration merged; this is not a dependency in the application's production runtime. Checked September 30, 2026.
  • NoKV — merged PR #518 adds a single-node test stack for LoopX's NoKV authority qualification. The author reports live conformance checks; the tool explicitly does not establish production availability, failover, or multi-owner operation. The separate metadata-runtime promotion gate issue #511 remains open. Its seed-routing helper PR #4724 is merged, while the companion ladder PR #4726 remains open; the helper merge alone does not close that promotion gate. Status: qualification tooling merged, with separate acceptance still open. This is concrete integration work beyond the README relationship below, not proof that NoKV runs LoopX in production. NoKV's current README evidence boundary separately marks full-surface native CLI and installed Python wheel acceptance against real services as not qualified. The historical single-node stack result does not establish those current, broader acceptances. Checked October 4, 2026.
  • GoTry (Danceiny) — issue #18, opened August 28, 2026, reports LoopX goals, Codex task bindings and initial heartbeat acknowledgements. A September 19 follow-up reports continued engineering automation and independent worktree delivery. PR #187 merged September 7; its reported local checks and successful exact-head CI concern a product delivery slice, whose body still lists real-inventory UAT as open. Merge and CI success do not close that acceptance. The October 2 checkpoint retains real-user evidence gates; issues #20, #22 and #142 remain open at review. Status: dated, author-reported development-workflow use. This does not establish a LoopX dependency in the travel application's runtime, independently reproduce the scheduler, or prove product acceptance. Checked October 5, 2026.
  • mimofan (XiaomingX) — PR #738, merged August 10, 2026, attributes UI/engine repairs to LoopX Todos. The changes fix /monitor persistence, /balance readback and /freeze false success; the PR's build/test results are author-reported. The checked main monitor implementation retains the application's own MonitorStore path. Status: dated, author-reported development workflow; UI fixes merged. This does not establish a LoopX dependency in mimofan's runtime; ongoing LoopX use, scheduler behavior and current-version compatibility were not independently verified. Checked October 5, 2026.
  • Meta-RLR (hk20013106) — PR #17, merged August 13, 2026, adds the external CLI/JSON maintenance boundary; scientific state remains owned by research_loop. Closed, unmerged PR #26 was superseded by PR #34, merged August 23, 2026. The checked main root entry point composes an outer maintenance adapter. Its activation and failure boundary requires explicit RLR_META_RLR_AUTOWAKE_CONFIG, classifies eligible failures and resolves verified repair provenance; the adapter preserves the original failure if maintenance is unavailable. The LoopX boundary uses external JSON CLI calls. Windows test jobs succeeded at PR #34's exact head; its continuous failure/repair/resume scenario remains an author-reported result. Status: optional maintenance and auto-wake integration merged. Current LoopX compatibility and sustained research use were not independently reproduced. Checked October 5, 2026.
  • LoopX Console (xielixing) — a third-party BitFun MiniApp for GitHub issue repair. Independent v3.9.29 was published August 18, 2026, with a MiniApp ZIP asset. The checked main worker calls the local LoopX CLI and quota should-run using the outer_controller profile; its fallback source installation pins LoopX v0.2.13. The same snapshot's README assigns scheduling and execution to the MiniApp/BitFun host and describes human approval before PR publication. Status: independently published, historical CLI integration observed. Release metadata and current source are separate evidence; the ZIP, real host execution, current LoopX compatibility and sustained use were not independently tested. OpenBitFun upstream inclusion remains a separate proposal below. Checked October 5, 2026.
  • zyra (BingruL) — the checked default branch is fix/execution-timeouts-and-diagnostics, pinned here to the September 19, 2026 source snapshot. Packaging includes the embedded LoopX modules and CLI entry points. Its source manifest pins LoopX v0.2.13; the listed source files match the upstream tag's tree. The runtime resolver requires the embedded source or a Zyra-owned installed distribution, without an archive fallback. Status: pinned source and packaging integration observed. No GitHub release or tag was listed at review; this does not establish whether a package was published elsewhere. Wheel installation, live Web/API execution, compatibility with current LoopX and sustained use were not independently verified. Checked October 5, 2026.
  • Hufu (Blicae8917) — PR #70, merged August 23, 2026, adds an opt-in LoopX v0.5.2 RunOnce Consumer. The checked main snapshot exports the consumer and its ports; its compatibility record retains the exact v0.5.2 baseline and no bundled LoopX dependency. The deployment provider supplies real transport and Host invocation. Issue #76 reports an owner-local pilot with committed/replayed RunOnce receipts, while general status projection remains incomplete; the issue is still open. Status: bounded integration merged, local pilot reported, companion projection open. That report was not independently reproduced and does not establish current-version compatibility or sustained deployment. Checked October 4, 2026.
  • benjamin-plugins (Yidada) — PR #1, merged September 5, 2026, adds a Codex plugin calling the official LoopX kernel. The checked main marketplace entry still registers the plugin. Its source record pins CLI qualification to a LoopX 0.5.4 checkout; the PR reports one independent model status scenario, with other scenarios structurally checked only. The preflight implementation locates an executable and reads registry shape without executing LoopX; it explicitly leaves runtime and driver verification false. Status: plugin merged, source-checkout validation reported. PyPI installation, actual background execution and current-version compatibility were not verified; scheduling remains host-owned. Checked October 4, 2026.
  • Adaptive-Agent-Orchestration-Protocol (YuemingHub) — PR #41, merged August 11, 2026, registers LoopX as an optional execution-continuity provider. The pinned pilot report reports Linux direct-CLI/custom-runner tests against LoopX v0.4.3: fresh-process recovery, validation and human gates, bounded handoff, accounting and rollback. The linked consumer Actions run succeeded at the recorded consumer head; this does not independently reproduce every reported assertion. The August 14 follow-up explicitly leaves LoopX unadopted, rather than promoting the pilot to ongoing adoption. AAOP is now archived; its retirement record, effective September 25, 2026, retains frozen research and releases without ongoing compatibility or support promises. Status: historical optional integration and bounded pilot in a retired project. Current LoopX compatibility, Windows/WSL and production host/session restart behavior remain unqualified; AAOP's retirement is not a demonstrated failure of LoopX. Checked October 5, 2026.

2. Mechanism Borrowing

These projects explicitly credit LoopX ideas. Native implementations and accepted design documents are distinct from depending on the LoopX runtime.

  • surogates (invergent-ai) — the pinned comparison and adoption plan selects durable grants, objective budgets and evaluator memory while retaining its own storage and runtime. PR #188, #190 and #191 merged August 3, 2026, implementing scoped approval grants, mission token allowances and prior-verdict memory. Current pinned MissionStore derives spend from its own session/task records; the evaluator checks the allowance and includes its previous verdict and stagnation count. PR #191 explicitly declines to port LoopX's progress-granularity lattice; the current stagnation prompt steers the judge rather than enforcing a verdict. Status: code-level borrowing; live PR status supersedes the plan's older table. The author reports passing focused PostgreSQL tests and 28 existing unit-suite failures; neither result was independently reproduced. This evidence does not qualify LoopX runtime integration, current compatibility or sustained use. Checked October 5, 2026.
  • future-os (futuregene) — PR #253 and #255, merged, implement selected multi-agent and goal-frontier mechanisms in Rust with explicit LoopX references. Status: native reimplementation.
  • gptme-contrib — PR #1373, merged, credits LoopX research for public/private evidence sanitization. Status: code-level borrowing.
  • KiroCrew — PR #3229, merged, adopts durable typed gates and debit-after-writeback in its perpetual-agent RFC. It deliberately retains different wake and work-selection mechanisms. Status: design adoption, documentation only.
  • multica (LRM-Teams) — PR #2174, merged, records five LoopX-inspired collaboration principles. Status: documentation only, with no LoopX runtime change.

3. Proposals and Deferred Adoption

  • MilkSU — issue #189 asks whether to adopt LoopX, borrow its durable Goal/state/retry ideas, or retain the existing ACP approach. Status: open design evaluation, not an accepted integration or implementation. Checked October 3, 2026.
  • ai-skills — issue #468 proposes LoopX-inspired quota-gated cooldown and self-wake checks without a LoopX runtime dependency. The September 23 follow-up leaves hook/upstream design work open because /loop is a built-in command. Status: historical open proposal, not a shipped skill or runtime adoption. Checked October 3, 2026.
  • OpenViking / VikingBot — open PR #5223 proposes optional, default-off LoopX-backed background long tasks. LoopX owns Goal/Todo state and execution gates; the Bot supplies the worker, model and tools. The author reports real CLI tests with a model substitute; real-model end-to-end acceptance and Docker build validation remain open. The proposal's pinned packaging places loopx==1.0.5 in the optional longtask extra; current main packaging does not declare that extra. These are proposal and main snapshots, not an instruction to install an unreleased extra. Status: runtime integration proposed, not merged or released. Checked October 4, 2026.
  • Opensiro VSM harness index — merged PR #607 records a feasibility stop for a frozen LoopX comparison; a subsequent registered-peer harness was built without live model execution. Merged PR #618 then retires that operated experiment under a public-evidence-only research policy. Status: historical research artifacts, live execution not performed or planned for that study; neither a new benchmark result nor runtime adoption. Checked September 30, 2026.
  • OpenBitFun (GCWing) — built-in console PR #2836 is open and replaces closed, unmerged #2382. The maintainer prioritizes beta stability before evaluating the larger feature. Status: upstream integration proposed, separate from the published third-party LoopX Console. Neither the independent release nor the author's reported local checks establishes an upstream merge or release. Checked October 5, 2026.
  • codexia — upstream PR #71 was closed unmerged after the author explained it targeted the wrong repository; downstream PR #1 remains open. Status: downstream proposal; the author explicitly reports that the CLI contract has not been checked against a live installation.
  • spoon-core — issue #285 proposes optional read-only LoopX control context before a model call. Status: open proposal.
  • GENesis-AGI — issue #2123 proposes evaluating LoopX before building a durable goal backend. Status: open evaluation request.
  • Orca — issue #12628 requests LoopX-like goal-driven iteration. Status: open user request, not maintainer acceptance or implementation evidence.
  • OpenAgentEmail — discussion #180 explores optional control-plane compatibility. The September 12 follow-up suggests a small reversible provider trial rather than critical-path adoption. That comment identifies itself as AI-authored. Status: discussion/pilot proposal.
  • Quesen — discussion #3735 led to a prepared-Effect risk-admission packet. It explicitly proposes shadow mode and preserves human authority. Status: integration packet, not a code proposal.
  • 8x8-user-edition — PR #63, merged, defers runtime adoption because of overlap with existing authority and state systems, while selecting protocol ideas. Status: runtime adoption deferred.
  • Mindthus — compare-and-absorb issue #132 is closed. Status: recorded evaluation; issue closure alone does not establish runtime adoption.
  • GovernLoop — Phase 0 capability-mapping PR #23 is closed unmerged. Status: historical evaluation proposal.
  • hartevo-desktop — issue #55 proposes a Mission Control kernel informed by Prime Agent and LoopX. Status: open design issue.
  • polyphemus — issue #89 studies LoopX primitives. Status: open research request.

4. Learning, Coverage and Collaboration

  • General Loop — its pinned README comparison contrasts LoopX's control-plane protocol with Markdown coordination on existing agent hosts. Status: comparative reference; the mention does not establish a LoopX dependency or adopted mechanism. Checked October 3, 2026.
  • GitHub-Michelin — merged PR #128 adds a bilingual LoopX editorial entry. Status: editorial coverage, not runtime adoption. Its September 27 assessment and snapshots are the editor's dated observations, not a live qualification of current LoopX behavior. Checked October 3, 2026.
  • ai-agent-book / Understanding AI Agents (bojieli) — PR #614, merged, introduces LoopX as a concrete Loop Engineering framework. Chapter 10 cites a fixed version and preserves its experimental evidence boundary. Status: teaching material, not reader adoption statistics.
  • NAVER fe-news — the September 2026 newsletter explains LoopX and its installation path in Korean. Status: editorial coverage, not a NAVER deployment claim.
  • OpenViking / NoKV — OpenViking's README lists LoopX; NoKV's README names an active open-source collaboration. Status: public project relationships; these listings alone do not establish a runtime dependency. Checked October 4, 2026.
  • loopx-book / loopx-book-labs (cocolord) — a bilingual, protocol-first developer book and runnable labs cover onboarding, issue-to-PR work and standalone extensions. Status: educational resources.

5. Derivatives

  • michaelx1993/loopx → foolzzz/loopx — one related fork family, not two independent adopters. The downstream changelog documents divergence from upstream and role-based orchestration. Merged michaelx1993 PR #27 introduces deterministic bookkeeping and a bounded orchestrator digest; merged foolzzz PR #18 prepares that fork's 2.0.0 release. Status: downstream code merged; these are fork changes, not an upstream LoopX release or independently verified performance gain. Checked October 3, 2026.
  • loopx-HPC (Sande33p) — PR #1 is merged in an independent fork, adding optional scientific campaigns, PBS/Slurm and MLflow integration. Status: downstream code merged; the author explicitly leaves live HPC acceptance pending.
  • foreman (needware) — PR #1 proposes a native TypeScript migration of the LoopX kernel pinned to an upstream commit. Status: open proposal, not a merged runtime migration.

Evidence Limits

  • A project's own development workflow, a packaged integration, protocol borrowing and public coverage are different relationships; do not add them together as a production-adopter count.
  • Stars, unchanged forks, automated Trending/digest posts and unrelated same-name projects are not adoption evidence.
  • Creator dogfooding and user-attributed showcases retain their own source boundaries in the Showcase catalog. For example, the MFS refactor case distinguishes publicly merged PRs from user-reported LoopX attribution.

Maintenance

  • Refresh both language versions through a pull request, checking source content, current PR merge state, replacement links and later comments. Record the date rather than treating an old label as current evidence.
  • Discovery queries include gh search code "huangruiteng/loopx", gh search issues loopx, gh search prs loopx and gh search repos loopx; review only public evidence and remove duplicate or unrelated matches.
  • Keep observed entries here; projects may voluntarily confirm their own use in ADOPTERS.md. Do not create self-attested entries on their behalf from this inventory.
  • Last reviewed: 2026-09-19. Public-source research: September 18; linked PR/issue statuses refreshed September 19.
  • Scoped update: 2026-09-30, covering CGC 2046, NoKV qualification tooling, the VikingBot proposal and the retired Opensiro study above. Other entries retain their earlier review boundary; this is not a full-table revalidation.
  • Scoped update: 2026-10-03, covering MilkSU, ai-skills, General Loop, GitHub-Michelin and the michaelx1993/foolzzz fork family. Current issue/PR states, later comments and pinned source files were checked; this update adds no confirmed runtime adopter and does not revalidate the rest of the table.
  • Scoped update: 2026-10-04–05, rechecking OpenViking, NoKV, Hufu and benjamin-plugins against current public PR/issue states and pinned source files. AAOP's historical pilot/retirement and LoopX Console's independent release, historical CLI pin, zyra's default branch and embedded source pin, and the separate OpenBitFun proposal were reconciled on October 5. GoTry's dated workflow reports were separated from its open product acceptance. Meta-RLR's merged auto-wake successor and current source wiring were reconciled with its historical PRs. mimofan's dated workflow attribution was separated from its application runtime. This distinguishes proposal packaging, historical tooling and pilots, plugin registration and current acceptance limits. No live integration was independently reproduced and other entries were not revalidated.