@ruvector/ruvllm-gpu-loop 0.0.0-stage → 1.1.0

package/GPU-LOOP.md

@@ -0,0 +1,238 @@
+# Scheduled GPU job loop
+
+Run a durable GPU job queue from the CLI, Claude Code or a Ruflo MCP host.
+The scheduler can be developed and contract-tested without live ruOS/Vast.ai
+access. A host adapter owns real provider calls and uses its existing ruOS
+authentication to resolve **secret references**, never export credential values.
+
+This release implements the schedule, durable state, reservations, host loader,
+file handoff, MCP tools and a live [Vast.ai host adapter](src/vast-host-adapter.mjs)
+([ADR 017](../../docs/adr/ADR-017-vast-host-adapter-and-live-paired-run.md)). That
+adapter has run paired evaluations and a training job end to end on rented GPUs,
+with local re-scoring and confirmed cleanup. It does not include a live ruOS
+connector or invent a secrets-service endpoint. No GPU rental or model download is
+started by installing the package. A zero-budget sample is disarmed.
+
+## Operator configuration
+
+Install the existing [integration](README.md), then prepare a private directory
+and copy [gpu-loop.example.json](examples/gpu-loop.example.json) outside your
+output directories. Keep the config owner-only (`chmod 600`); create its
+`stateRoot` with mode 700. Use canonical absolute paths without symlinks.
+
+Edit the example's paths, model pin, resource ID and `secretRef`. The example
+secret alias and instance ID are placeholders, not discovered ruOS resources.
+Keep `budgetUsd: 0` until a spending ceiling is explicitly chosen. Merely reading
+status does not submit work. A zero-budget tick records a budget-limit failure
+and never calls the provider. Use a new state directory for a new reviewed
+configuration; state binds the config's exact byte digest.
+
+Configuration fields:
+
+| Field | Meaning |
+| --- | --- |
+| `intervalSeconds` | Polling cadence, 1–86400 seconds; active deadlines and pending cleanup can become due sooner |
+| `maxAttempts` | Maximum attempts per job, 1–10; retries require confirmed cleanup |
+| `budgetUsd` | Total conservative admission budget in exact microdollars; zero disarms submissions |
+| `jobs` | 1–128 fixed operator-approved jobs |
+| `model` | Exact `repo` and 40-character model revision |
+| `task` | `evaluate` or `train`; the host must implement that workload |
+| `secretRef` | Opaque alias resolved only by the authenticated host |
+| `resource` | Exactly one string `instanceId` or `offerId`; no automatic offer selection |
+| `maxCostUsd` | Reserved cost ceiling for each attempt; never refunded automatically |
+| `timeoutSeconds` | Per-attempt deadline, 1–86400 seconds |
+| `experiment` | Optional absolute `configPath`/`planPath` for the host to execute using the native experiment workflow |
+
+The schedule advances a finite queue. It does not reset succeeded jobs or start
+unlimited training cycles. Another experiment cycle needs a new reviewed config
+and state root. Reservations limit admitted work; they cannot impose a hard cap
+on Vast billing. The host must separately enforce actual spend and worker runtime.
+
+## Host adapter
+
+The host module must be a canonical, owner-only `.mjs` file pinned by SHA256.
+Its entire transitive dependency tree is trusted operator code. No provider effects
+or secret logging may occur at import: all effects belong in the methods below.
+This interface is supplied by RuVLLM; these names are not claimed ruOS API methods.
+
+```js
+export async function submit({ job, idempotencyKey, signal }) {
+  // Resolve job.secretRef at the authenticated host. Submit OR find the exact
+  // existing job by idempotencyKey. Enforce timeout/cost before starting work.
+  // Return only a bounded provider run identifier, never credential material.
+  return { runId };
+}
+export async function inspect({ job, runId, idempotencyKey, signal }) {
+  // Read the real executor's structured evidence. Do not trust worker stdout.
+  return { status: 'running' }; // Or failed/completed with costUsd and completion.
+}
+export async function cleanup({ job, runId, signal }) {
+  // Stop/release only this job's allocation. Do not destroy a shared instance.
+  // Confirm absence/release through the provider before returning true.
+  return { released: false };
+}
+```
+
+The snippet is an interface illustration, not a runnable provider implementation.
+Use the host's actual supported ruOS secrets and Vast execution routes. Honor
+AbortSignal and implement idempotent submit/cleanup. A thrown/late/ambiguous submit
+blocks automatic resubmission, since a provider may already have allocated a GPU.
+
+A completed result needs this structured contract, derived by the trusted host:
+
+```json
+{
+  "status": "completed",
+  "costUsd": 0.25,
+  "completion": {
+    "completionVersion": 1,
+    "completionVerified": true,
+    "status": "ok",
+    "exitCode": 0,
+    "assertionsVerified": true,
+    "evidenceComplete": true,
+    "truncated": false,
+    "jobId": "EXACT_JOB_ID",
+    "modelRepo": "Qwen/Qwen3.5-4B",
+    "modelRevision": "851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a",
+    "runId": "EXACT_CURRENT_RUN_ID",
+    "idempotencyKey": "EXACT_CURRENT_ATTEMPT_KEY"
+  }
+}
+```
+
+Boolean flags are structural checks, not cryptographic attestation. Bind evidence
+to the actual run and retain raw measurements at the host. An old attempt's reply,
+wrong model, missing assertions or unverified completion cannot mark a job
+successful. Terminal success also requires explicit resource-release confirmation.
+
+## Concrete file handoff
+
+For a ruOS-side scheduled consumer, use the bundled
+`src/gpu-spool-adapter.mjs`. Set `RUVLLM_GPU_SPOOL_ROOT` to an existing private
+canonical directory. This adapter performs **no provider calls**; it hands work
+to a host sharing that directory, or a securely synchronized copy of the files.
+Transfer and remote authentication are separate host responsibilities.
+
+For each persisted attempt key, files are:
+
+| File | Writer and schema |
+| --- | --- |
+| `KEY.request.json` | Scheduler: `ruvllm-gpu-spool-request/v1`, `runId`, `job`, credential-handling notice |
+| `KEY.status.json` | Host: `ruvllm-gpu-spool-status/v1`, matching `runId`, status, optional cost/completion |
+| `KEY.cleanup-request.json` | Scheduler: `ruvllm-gpu-spool-cleanup-request/v1`, `runId`, `jobId` |
+| `KEY.cleanup.json` | Host: `ruvllm-gpu-spool-cleanup/v1`, matching `runId`, boolean `released` |
+
+All files must be owner-only, regular, nonsymlink files of at most 64 KiB. The
+host should publish replies atomically. Repeated identical requests are idempotent;
+conflicting contents under the same key are rejected. No status reply means
+pending/running, not success. Missing cleanup acknowledgement keeps cleanup
+pending. The request's `secretRef` is an alias; no credential belongs in any file.
+
+Example local bindings (operator paths, not secrets):
+
+```sh
+export RUVLLM_GPU_LOOP_CONFIG=/absolute/private/gpu-loop.json
+export RUVLLM_GPU_SPOOL_ROOT=/absolute/private/gpu-spool
+export RUVLLM_GPU_HOST_MODULE="$PWD/integrations/metaharness/src/gpu-spool-adapter.mjs"
+chmod 600 "$RUVLLM_GPU_HOST_MODULE"
+export RUVLLM_GPU_HOST_SHA256="$(shasum -a 256 "$RUVLLM_GPU_HOST_MODULE" | awk '{print $1}')"
+```
+
+Prepare the private spool/state directories and config before ticking. The
+source module's chmod is needed after a normal Git checkout because host modules
+are deliberately required to be owner-only. A custom trusted module can replace
+the spool adapter; recalculate and review its pin after any code change.
+
+## Vast.ai host adapter
+
+`src/vast-host-adapter.mjs` is a real provider adapter (see
+[ADR 017](../../docs/adr/ADR-017-vast-host-adapter-and-live-paired-run.md)). Set
+`RUVLLM_VAST_HOST_CONFIG` to an owner-only `ruvllm-vast-host/v1` file:
+
+| Field | Meaning |
+| --- | --- |
+| `apiBase` | `https://console.vast.ai` |
+| `secrets` | `{ "<secretRef>": { "file": "/abs/owner-only/key" } }`; the key is sent only as a Bearer header |
+| `image` | Container image pinned by digest |
+| `repoUrl`, `commit`, `bundle` | Public checkout and exact commit the rental runs; frozen bundle name |
+| `diskGb`, `armDeadlineSeconds` | Rental disk; per-arm worker deadline |
+| `guardGraceSeconds` | Extra time after `timeoutSeconds` before the independent destroy timer fires |
+| `resultsRoot`, `sshKey`, `repoRoot`, `python`, `node` | Absolute local paths for receipts, retrieval and local re-scoring |
+| `sshKnownHosts` | Absolute path to an owner-managed `known_hosts` file (not group/world-writable) holding the pinned host keys; ssh runs with `StrictHostKeyChecking=yes` and refuses unknown or changed keys |
+| `sshTrustOnFirstUse` | `true` to accept a rental's first host key into a fresh per-instance file (the pre-1.1.0 behaviour). One of `sshKnownHosts` or `sshTrustOnFirstUse: true` is required; the config is rejected (`VAST_HOST_CONFIG_SSH_TRUST`) before anything is rented otherwise |
+| `maxEvidenceBytes`, `maxEvidenceEntries` | Optional caps on the retrieved evidence archive (defaults 2 GiB and 20000 entries) |
+
+Jobs must use `resource.offerId`. Submit refuses the offer when its hourly price
+times `timeoutSeconds` exceeds `maxCostUsd`, labels the rental with the attempt's
+idempotency key, and arms a `systemd-run --user` destroy timer. Inspect reports
+accrued cost, pulls raw outputs when the remote run is done, and re-scores them
+locally. The pull stops at `maxEvidenceBytes`, and every tar header is checked
+before extraction: only regular files and directories, only relative paths under
+`out/` or `run/` with no `..`, within the entry and size caps. A rejected archive
+is never extracted, fails the attempt and is recorded in `evidence-rejected.json`; only an identical report counts as verified completion. Cleanup returns
+`released: true` only after the instance is absent from the account listing.
+Receipts (`submission.json`, `progress.json`, `verification.json`,
+`cleanup-ID.json`, `guard-ID.json`) contain no credential. Copy the module to a
+private owner-only path and pin that copy's SHA256, as for any host module.
+
+## CLI and loop schedule
+
+```sh
+node integrations/metaharness/src/gpu-loop-cli.mjs status --config "$RUVLLM_GPU_LOOP_CONFIG"
+node integrations/metaharness/src/gpu-loop-cli.mjs tick \
+  --config "$RUVLLM_GPU_LOOP_CONFIG" \
+  --adapter "$RUVLLM_GPU_HOST_MODULE" --adapter-sha256 "$RUVLLM_GPU_HOST_SHA256"
+node integrations/metaharness/src/gpu-loop-cli.mjs run \
+  --config "$RUVLLM_GPU_LOOP_CONFIG" \
+  --adapter "$RUVLLM_GPU_HOST_MODULE" --adapter-sha256 "$RUVLLM_GPU_HOST_SHA256" \
+  --max-ticks 60
+```
+
+`run` defaults to one tick, caps the count at 10000, and waits until the next due
+poll/deadline/cleanup step. SIGINT/SIGTERM stops between ticks; it does not erase
+running work. State survives restarts. Alternatively, have the operator's existing
+scheduler invoke `tick`; overlapping processes refuse the exclusive lock. A coarse
+external interval can delay cleanup, so the host must always enforce its own
+deadline. No cron entry or background process is installed automatically.
+
+## Claude Code and Ruflo
+
+Load [ruflo-metaharness](../../plugins/ruflo-metaharness/README.md) with the four
+bindings above in its process environment. `gpu_loop_status` and `gpu_loop_tick`
+take `{}` only. Configuration, host code and budget cannot come from tool args.
+The `/ruflo-metaharness:gpu-loop` command checks status and advances a tick within
+the operator's configured scope. Ruflo can use the same MCP server and bindings.
+
+If the Claude host supports recurring `/loop`, it can repeatedly invoke this command:
+
+```text
+/loop 1m /ruflo-metaharness:gpu-loop
+```
+
+That host-specific facility was not E2E-tested here. The durable scheduler still
+enforces its own due time, budget, lock and submission state across host restarts.
+Existing native experiment tools remain separate; no schedule grants activation.
+
+## Reconciliation and validation
+
+`blocked` means the submit outcome is uncertain. `cleanup` means resource release
+has not yet been confirmed. Do not delete state or launch a replacement rental
+to make either status disappear. Use the saved key/run ID to inspect the actual
+host/provider, reconcile its resources, and only then repair state or prepare a
+new reviewed cycle. A crash may leave `tick.lock`; it is never auto-broken.
+
+Run the local scheduler/host/spool tests and all plugin wire tests through the
+existing `npm test`/Node commands. Test adapters and host files are explicitly
+mock contracts; no secrets, provider, model inference or GPU E2E is exercised.
+The [ADR](../../docs/adr/ADR-015-scheduled-gpu-loop-and-host-secret-boundary.md)
+records the state machine, failure paths and threat boundary.
+
+Validation on 2026-11-03: **60 local tests passed**, with no failures or skips:
+17 scheduler tests, 8 host/CLI tests, 3 spool tests (including the scheduled
+handoff), 3 new GPU MCP wire tests, and the original 29 integration/plugin tests.
+The long-interval deadline test uses a 60-second cadence and a one-second job
+deadline and confirms cleanup becomes due without waiting the full cadence.
+Repository integrity and unchanged native-source provenance checks also passed.
+Live ruOS/Vast.ai, actual Claude `/loop`, secrets resolution and GPU inference/
+training remain deliberately unrun. Existing CI test globs include these tests.

package/LICENSE

@@ -0,0 +1,201 @@
+Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

package/NOTICE

@@ -0,0 +1,9 @@
+RuFlo Code Model
+Copyright 2026 rUv and contributors
+
+Original experiment code and synthetic tool traces are licensed under Apache 2.0.
+Base model: Qwen/Qwen3.5-4B, Apache 2.0, owned by the Qwen team.
+HumanEval data in data/external is MIT licensed; see HumanEval.LICENSE.
+MLX LM, Transformers, PEFT, and Claude Code Router retain their respective licenses.
+Claude Code is a separate proprietary client. No Claude model weights or private
+Claude training data are included. This project is not endorsed by Anthropic.

package/README.md

@@ -1,3 +1,205 @@
-# Temporary Holding Version
+# MetaHarness weight experiments
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+<p align="center"><img src="https://raw.githubusercontent.com/ruvnet/ruvllm/ruvllm-npm-v4.0.1/docs/assets/gpu-job-loop.svg" alt="The GPU job loop checks a spending ceiling, rents the offer and arms a destroy timer, runs the pinned job, pulls raw outputs, checks them locally, then destroys the rental and confirms it is gone." width="100%"></p>
+
+<p align="center"><img src="https://raw.githubusercontent.com/ruvnet/ruvllm/ruvllm-npm-v4.0.1/docs/assets/heldout-results.svg" alt="Held-out results on 256 untouched tasks: tool passes 149, 176 and 192 of 192 for base, coding-4b and coding-4b-r2; arithmetic 62, 64 and 46 of 64. coding-4b is kept; coding-4b-r2 is rejected." width="100%"></p>
+
+Published on npm as [`@ruvector/ruvllm-gpu-loop`](https://www.npmjs.com/package/@ruvector/ruvllm-gpu-loop).
+The package ships the JavaScript controller, GPU loop and Vast.ai host adapter.
+It does not ship the native `weight_evolution` binary, model weights or
+adapters. Build the binary from this repository and pass its path with
+`--binary`.
+
+```sh
+npm install @ruvector/ruvllm-gpu-loop
+npx ruvllm-gpu --help        # bounded weight-edit experiments
+npx ruvllm-gpu-loop status --config /abs/private/gpu-loop.json
+```
+
+### Package name: `@ruvector/ruvllm-gpu-loop`, not `@ruvector/ruvllm`
+
+Versions 4.0.0 and 4.0.1 of this harness were published as `@ruvector/ruvllm`
+by mistake. That name belongs to the RuVLLM inference runtime (GGUF inference,
+TurboQuant KV cache, SONA, FlashAttention) built from
+[`ruvnet/RuVector`](https://github.com/ruvnet/RuVector/tree/main/npm/packages/ruvllm),
+and the 4.0.x upload replaced it for everyone. `@ruvector/ruvllm@4.1.0`
+restores the runtime and depends on this package, so its GPU-loop subpaths keep
+working:
+
+| 4.0.x (`@ruvector/ruvllm`) | Now (`@ruvector/ruvllm-gpu-loop`) | Via `@ruvector/ruvllm@^4.1.0` |
+| :--- | :--- | :--- |
+| `import … from '@ruvector/ruvllm'` (`runExperiment`) | `import … from '@ruvector/ruvllm-gpu-loop'` | `@ruvector/ruvllm/engine` |
+| `@ruvector/ruvllm/gpu-loop`, `/adapters`, `/gpu-host`, `/gpu-spool-adapter`, `/vast-host-adapter`, `/fixture` | same subpaths on `@ruvector/ruvllm-gpu-loop` | same subpaths |
+| `npx ruvllm run …` / `npx ruvllm fixture …` | `npx ruvllm-gpu run …` / `npx ruvllm-gpu fixture …` | `npx ruvllm gpu run …` |
+| `npx ruvllm-gpu-loop`, `npx ruvllm-adapters` | unchanged | unchanged |
+
+A `prepublishOnly` guard (`scripts/check-package-name.mjs`) refuses to publish
+this package under the runtime's name.
+
+Run multiple bounded RuVLLM edits against one unchanged checkpoint, validate
+supplied evaluations, package RVF evidence, restore every patched checkpoint,
+and rank research-qualified candidates through MetaHarness. This controller
+uses the actual Flywheel, Darwin and Weight-EFT packages pinned in
+[package-lock.json](package-lock.json).
+
+The runner does not perform inference, discover an effective weight edit, train
+an adapter or activate a model. CPU fixtures exercise real file editing and
+restoration with explicitly synthetic evaluation receipts. External evaluation
+files remain caller-supplied evidence; hashes do not authenticate their origin.
+
+## Install and run the CPU fixture
+
+Requirements: Linux or macOS, Node 22+, Python 3.12+, a C linker, and Rust 1.97.1
+as pinned by `native/rust-toolchain.toml`. Activate any locally installed Rust
+toolchain before using Cargo. From the repository root:
+
+```sh
+(cd integrations/metaharness && npm ci --ignore-scripts)
+(cd native && cargo build --locked --release --bin weight_evolution)
+python3 native/verify_source.py
+node integrations/metaharness/src/cli.mjs fixture \
+  --dir "$PWD/runtime/metaharness-fixture" \
+  --binary "$PWD/native/target/release/weight_evolution"
+```
+
+Create the fixture's parent directory first (`mkdir -p runtime`). Its own directory
+must not exist: exclusive creation prevents accidental replacement. For another
+fixture use a new directory. The report should select `fixture-improvement`, leave
+`unevaluated-control` unqualified, verify RVF and exact rollback for both, and
+report `inferencePerformed: false`, `independentConfirmation: false` and
+`productionActivation: false`. A selected synthetic candidate says nothing about
+actual model accuracy.
+
+## Configure real experiment inputs
+
+Prepare reviewed source checkpoint, patch, policy and optional evaluation JSON
+under explicit input roots. Native request/data contracts are described in the
+[native guide](../../native/README.md). Patch generation remains model specific;
+the controller does not guess useful tensor edits.
+
+Run the project-local installer with paths you control:
+
+```sh
+bash scripts/install-metaharness.sh \
+  --config-dir /absolute/private/operator-profile \
+  --read-root /absolute/checkpoints \
+  --read-root /absolute/experiment-inputs \
+  --policy /absolute/experiment-inputs/policy.json \
+  --output-root /absolute/private/experiment-outputs
+```
+
+It installs frozen npm dependencies without lifecycle scripts, builds the native
+binary, verifies native provenance and runs its smoke fixture. It creates a
+private runtime profile and Claude launcher without changing global settings.
+Repeat `--read-root`/`--policy` for additional approved inputs. Existing profile
+files are never replaced; use another directory for a new profile. Existing config
+and output directories must be owner-only. `--configure-only` skips installation
+and building but still checks native readiness.
+
+Example plan inside an approved read root:
+
+```json
+{
+  "schema": "ruvllm-experiment-plan/v1",
+  "evidenceMode": "external",
+  "source": "/absolute/checkpoints/source.safetensors",
+  "policy": "/absolute/experiment-inputs/policy.json",
+  "candidates": [
+    {
+      "id": "candidate-01",
+      "patch": "/absolute/experiment-inputs/patch.json",
+      "evaluation": "/absolute/experiment-inputs/evaluation.json"
+    }
+  ]
+}
+```
+
+```sh
+node integrations/metaharness/src/cli.mjs run \
+  --config /absolute/private/operator-profile/runtime.json \
+  --plan /absolute/experiment-inputs/plan.json
+```
+
+Omit `evaluation` to run mechanical checks without qualification. After an external
+inference runner has evaluated a candidate, submit its hash-bound evaluation in a
+new plan/run. Candidate checkpoint hashes are deterministic for the same patch;
+new artifact paths do not relax identity checks. Paired baseline, protocol, suite,
+runtime, precision and output-budget consistency remain the evaluator's responsibility.
+
+Runs use new private `run-*` directories and preserve rejected candidates. Reports
+bind config/plan/source/policy/binary/gate/patch/evaluation identities and record
+rollback and RVF verification. Native rejection records a reason. Structural or
+restoration failures abort; interrupted runs can leave artifacts. Reconcile those
+directories before retrying; never erase failed evidence to manufacture success.
+
+## Claude Code and Ruflo
+
+Load [ruflo-metaharness](../../plugins/ruflo-metaharness/README.md) using the
+generated `launch-claude.sh`, or set `RUVLLM_EXPERIMENT_CONFIG` and use Claude's
+`--plugin-dir`. The `/ruflo-metaharness:weight-experiment` command and reviewer
+agent guide the bounded workflow. Its single MCP tool accepts a plan path; the
+model cannot override the operator configuration. The same stdio server can be
+registered in a Ruflo project's MCP configuration. Portable wire tests exercise
+the real engine; installed Claude/Ruflo client acceptance is separate.
+
+## Scheduled GPU work
+
+For scheduled GPU work, use the [durable GPU loop](GPU-LOOP.md). It supports
+host-owned ruOS secret references, Vast job descriptors, persisted reservations,
+deadline-aware polling and confirmed cleanup. Local contract tests require no
+live provider. The bundled file handoff keeps requests pending for a trusted host.
+
+## Weight-EFT data and plans
+
+The data-only adapter CLI reads JSON stdin, with a 16 MiB bound:
+
+```sh
+node integrations/metaharness/src/adapters-cli.mjs export < reviewed-archive-input.json > export.json
+node integrations/metaharness/src/adapters-cli.mjs training-plan < training-input.json > plan.json
+node integrations/metaharness/src/adapters-cli.mjs select < measured-candidates.json > selection.json
+```
+
+`export` takes `{archive, evalHoldout, maxTokens?}`. Archive entries follow the
+pinned Weight-EFT `DarwinTrajectory` contract (`instance_id`, `model`, `tier`,
+boolean `resolved`, `messages`, `model_patch`). All actual evaluation instance IDs
+must be in `evalHoldout`. Output contains standard `sft`/`dpo` arrays, drop reports
+and input hashes. Reward-hacking filtering stays enabled, overlength traces are
+dropped, and tool calls are preserved. These filters cannot authenticate outcomes
+or detect all semantic contamination.
+
+`training-plan` takes `{base:{id,paramsB},sftPath,dpoPath,adapterPrefix}`. It emits
+upstream plans plus `executionStatus: not_executed` and
+`runnerCompatibility: adapter-required`. This repository has no upstream
+`ruvllm microlora` executable. Its existing fixed CUDA/PEFT experiment runner is
+not a configurable SFT/DPO trainer. No GPU, model download or paid job is started.
+
+`select` takes `{baseline,candidates,envelope}` with entries
+`{id,score,evidenceSha256}`. It replays finite, supplied scores through Darwin's
+contrastive engine. It is development ranking with an immutable envelope and no
+independent confirmation; it cannot authenticate caller-supplied evidence. The
+native runner applies its stricter qualification first.
+
+## Tests and benchmark
+
+```sh
+(cd integrations/metaharness && npm test)
+node --test plugins/ruflo-metaharness/test/*.test.mjs
+node integrations/metaharness/bench/run.mjs --iterations 7 --warmup 2 --candidates 2
+```
+
+Tests require the built native binary. In restricted agent environments the Node
+test runner, native child processes and package fetching may require the host's
+supported execution permission. Verify actual subtest counts; a file-only runner
+summary is insufficient evidence. Do not disable TLS/checksums or native checks.
+
+The benchmark compares fresh versus reused Node processes on identical CPU
+fixtures, retaining raw latency samples and implementation/binary/lock hashes.
+The reused process matches the MCP server architecture. All native hashing,
+patch validation and byte restoration remain enabled. See the
+[benchmark report](../../docs/benchmarks/metaharness-cpu.md) for measured scope
+and limits; no model-performance claim follows from orchestration speed.
+
+Architecture and threat boundaries are recorded in
+[ADR 012](../../docs/adr/ADR-012-metaharness-bounded-experiment-runner.md),
+[ADR 013](../../docs/adr/ADR-013-training-and-development-selection.md) and
+[ADR 014](../../docs/adr/ADR-014-claude-ruflo-integration.md).