docs: accuracy fixes and copy pruning by soleil-colza · Pull Request #203 · arkorlab/arkor

soleil-colza · 2026-07-24T08:38:28Z

Summary

Documentation-only pass over the reference pages, in three commits (English and Japanese kept in lockstep throughout):

Prune positioning copy and correct stale claims

The introduction drops its three essay sections (who-it's-for pitch, Python-ecosystem relations, tagline expansion) in favor of one factual lead sentence.
Removes the cookbook thesis paragraph, models' "why one model" justification, and the repeated "fast and stable" marketing tails in models/roadmap, plus motivational hooks opening four recipes and assorted tagline clauses.
Stale claims fixed: bring-your-own datasets work today (Hugging Face name or hosted file URL) and are now listed under what-works; the redaction template redacts rather than extracts; the cookbook's not-covered bullet no longer implies production serving is unavailable (deployments exist — only the recipes are missing).
models.mdx no longer claims createTrainer accepts a single model: the one-model list is the Playground/hosted-inference surface, while training accepts any loadable identifier.

Point production readers at deployments, fix Playground claims

Studio's "when not to use" sections (concepts + studio overview) directed production traffic at infer, which only exists inside onCheckpoint during a run — they now point at publishing a deployment and calling its OpenAI-compatible endpoint.
The dev guide claimed the Playground can chat with checkpoints as they land; it lists completed runs only (mid-run checkpoints are code-only via onCheckpoint({ infer })).
The Playground page suggested editing src/arkor/trainer.ts to test a system prompt, but the Playground builds its request solely from the page's inputs; the working path is programmatic infer with a system message.
mid-run-eval's "compare against the base model" variation could not do what it promised (the callback's infer is bound to the checkpoint, InferArgs has no base-model override) — reframed as forwarding checkpoint samples to a review queue.

Fix npm tab commands and punctuation-stripped anchors

The npm package-manager tabs invoked the nonexistent npm arkor ... form; now npx arkor ... (26 occurrences).
Several fragment links kept =, /, or full-width-paren characters that GitHub-style heading-id sluggers strip, so they never jumped to their section.
Removes the hosted-inference-endpoint roadmap card: stable per-model HTTPS endpoints shipped as deployments.

Summary by cubic

Fixes inaccurate docs and broken examples, trims marketing copy, and points production users to deployments. Incorporates review feedback by restoring tab/anchor conventions and tightening wording; addresses ENG-955.

Bug Fixes
- Restore conventions: npm arkor ... in package-manager tabs and Mintlify anchor fragments (preserve =, /, full-width parens).
- Production guidance and security: direct users to publish a deployment and call its OpenAI-compatible endpoint; clarify Studio runs on loopback but remains token-gated (not “no inbound connection”).
- Playground/dev guide: lists completed runs only; no system-prompt field—use onCheckpoint’s infer and pass a system message.
- Mid-run eval/structured outputs: note infer is bound to the checkpoint; for base-vs-adapter comparisons use CloudApiClient.chat({ baseModel }); add maxTokens truncation caveat.
- Correct feature claims: bring-your-own datasets work (Hugging Face name or hosted file URL); redaction template redacts; roadmap/models pages drop the “single-valued model field” wording and remove the hosted-endpoint card (deployments shipped).
Refactors
- Prune positioning copy to a single factual lead; tighten phrasing (callbacks “invoke the functions you register”); keep English and Japanese in lockstep.
- Clarify model support: Playground/hosted inference serve gemma-4-E4B-it; createTrainer.model accepts any backend-loadable model identifier.

^{Written for commit ed8640d. Summary will update on new commits.}

Summary by CodeRabbit

Documentation
- Clarified lifecycle callbacks, checkpoint evaluation, notifications, structured outputs, and early-stopping guidance.
- Updated Studio guidance to recommend deployments and OpenAI-compatible endpoints for production use.
- Expanded introduction and CLI documentation for Studio, Playground, training runs, and supported workflows.
- Clarified supported models and refined roadmap and cookbook coverage in English and Japanese documentation.

Remove essay/positioning material that doesn't belong in reference docs (the who-it's-for pitch, the Python-ecosystem essay, and the tagline section in the introduction; the cookbook thesis paragraph; models' why-one-model justification; repeated marketing tails in models/roadmap; assorted tagline clauses and recipe hooks), and fix stale claims: bring-your-own datasets work today (HF name or hosted file URL), the redaction template redacts rather than extracts, and production serving exists via deployments — the cookbook bullet now says only the recipes are missing.

Studio's when-not-to-use sections told production readers to call infer, which only exists inside onCheckpoint during a run — they now point at publishing a deployment and calling its OpenAI-compatible endpoint. The dev guide claimed the Playground chats with checkpoints as they land (it lists completed runs; mid-run checkpoints are code-only via onCheckpoint), and the Playground page suggested editing src/arkor/trainer.ts to test a system prompt even though the request is built solely from the page's inputs.

The npm package-manager tabs invoked the nonexistent 'npm arkor ...' form (npm has no bin passthrough) — now 'npx arkor ...' — and several fragment links kept '=' / '/' / full-width-paren characters that GitHub-style heading-id sluggers strip, so they never jumped to their target section. Also removes the hosted-inference-endpoint roadmap card: stable per-model HTTPS endpoints shipped as deployments.

qodo-code-review · 2026-07-24T08:38:35Z

ⓘ Qodo reviews are paused because your trial has ended. Ask your workspace admin to add credits to resume reviews. Manage billing

coderabbitai · 2026-07-24T08:38:46Z

Walkthrough

Documentation updates align English and Japanese guidance for lifecycle callbacks, Studio boundaries, checkpoint inference, cookbook patterns, supported models, current capabilities, and roadmap status.

Changes

Documentation updates

Layer / File(s)	Summary
Lifecycle and Studio guidance `docs/concepts/`, `docs/ja/concepts/`, `docs/guides/cli/dev.mdx`, `docs/studio/`, `docs/ja/studio/`	Lifecycle callback descriptions, Studio production guidance, checkpoint inference access, and system-prompt testing instructions were revised.
Cookbook explanations `docs/cookbook/`, `docs/ja/cookbook/`	Cookbook content now describes checkpoint-bound inference, review workflows, notifications, early stopping, structured outputs, and serving coverage.
Product overview and roadmap `docs/introduction.mdx`, `docs/ja/introduction.mdx`, `docs/models.mdx`, `docs/ja/models.mdx`, `docs/roadmap.mdx`, `docs/ja/roadmap.mdx`	Current Studio, training, dataset, hosted model, deployment, and roadmap descriptions were updated.
Development workflow guidance `docs/guides/cli/dev.mdx`	Post-training Playground usage and programmatic access to mid-run checkpoints were clarified.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: copilot, k-taro56

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the documentation accuracy updates and copy pruning across the PR.

✨ Finishing Touches

📝 Generate docstrings

Create stacked PR
Commit on current branch

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch eng-955

✨ Simplify code

Create PR with simplified code
Commit simplified code in branch eng-955

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

drift-check · 2026-07-24T08:39:28Z

Code Review Bot

No comment/code divergences detected. Reviewed 26 file(s); skipped 1.

github-code-quality · 2026-07-24T08:40:15Z

Code Coverage Overview

Languages: TypeScript

TypeScript / code-coverage/arkor

The overall coverage in commit ed8640d in the eng-955 branch remains at 98%, unchanged from commit 497243d in the main branch.

TypeScript / code-coverage/create-arkor

The overall coverage in commit ed8640d in the eng-955 branch remains at 60%, unchanged from commit 497243d in the main branch.

TypeScript / code-coverage/cli-internal

The overall coverage in commit ed8640d in the eng-955 branch remains at 97%, unchanged from commit 497243d in the main branch.

TypeScript / code-coverage/studio-app

The overall coverage in commit ed8640d in the eng-955 branch remains at 53%, unchanged from commit 497243d in the main branch.

_{Updated July 25, 2026 10:00 UTC}

codecov · 2026-07-24T08:40:22Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5790d7a7ee

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-07-24T08:41:50Z


 ```bash npm
-npm arkor login
+npx arkor login


Restore the prescribed npm command form

Keep the npm tab as npm arkor login; this repository deliberately standardizes unscripted CLI examples under /cli/* and /guides/cli/* on the five-tab <pm> arkor <subcommand> form. The same npx rewrite is repeated for init/logout/whoami and in the Japanese mirrors, making this documentation inconsistent with the project-wide convention.

AGENTS.md reference: AGENTS.md:L109-L109

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-07-24T08:41:50Z

+| `CI=1` (or any non-TTY stdout) | Force non-interactive mode. `arkor init` skips its prompts and uses the values you passed via flags (or built-in defaults under `--yes`); git init runs without asking when `--git` or `--yes` is set, and is skipped silently otherwise. See [`arkor init` § CI / non-interactive shells](/cli/init#ci--non-interactive-shells). |
+| `CLAUDECODE=1` | Strict mode for AI agents that spawn the CLI without being able to answer interactive prompts. `arkor init` (and `create-arkor`) refuse to run unless every curated flag is present, printing a multi-line stderr block with the suggested re-invocation; pass `-y`/`--yes` to opt back into "accept all defaults". See [`arkor init` § Claude Code (`CLAUDECODE=1`) strict mode](/cli/init#claude-code-claudecode1-strict-mode). |


Restore Mintlify’s actual heading fragments

Restore these fragments to #ci-/-non-interactive-shells and #claude-code-claudecode=1-strict-mode: Mintlify preserves / and = in these rendered IDs, so both new links miss their headings. The same incorrect normalization also breaks the edited quickstart/init cross-links and the Japanese links where full-width parentheses are preserved.

AGENTS.md reference: AGENTS.md:L110-L116

Useful? React with 👍 / 👎.

greptile-apps · 2026-07-24T08:42:12Z

Greptile Summary

Documentation-only accuracy and copy-editing update.

Corrects Studio, Playground, deployment, dataset, model, and template guidance.
Fixes CLI examples and broken fragment links.
Prunes marketing copy while keeping English and Japanese documentation aligned.

Confidence Score: 5/5

The documentation-only PR appears safe to merge.

No blocking failures remain.

Important Files Changed

Filename	Overview
docs/concepts/studio.mdx	Corrects production deployment guidance and removes the previously reported prohibited punctuation.
docs/cookbook/mid-run-eval.mdx	Replaces an unsupported base-model comparison example with checkpoint sample review guidance.
docs/introduction.mdx	Condenses introductory copy and updates current dataset and template capabilities.
docs/models.mdx	Clarifies the distinction between hosted inference model support and trainer model identifiers.
docs/studio/playground.mdx	Corrects Playground checkpoint and system-prompt instructions.
docs/ja/introduction.mdx	Mirrors the English introduction and capability corrections in Japanese.

_{Reviews (2): Last reviewed commit: "Merge remote-tracking branch 'origin/mai..." | Re-trigger Greptile}

coderabbitai

Actionable comments posted: 8

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/cli/auth.mdx`:
- Line 27: Restore the required npm arkor command form in all affected CLI
CodeGroups: docs/cli/auth.mdx lines 27, 93, and 141; docs/cli/init.mdx lines 23,
135, 204, 230, and 256; and docs/guides/cli/auth.mdx lines 19, 47, and 75.
Replace each npx invocation with the corresponding npm arkor subcommand while
preserving its existing arguments and examples.

In `@docs/concepts/studio.mdx`:
- Line 27: Clarify the loopback security wording in docs/concepts/studio.mdx at
lines 27-27 to state that there is no inbound access from outside the machine,
while preserving that local loopback connections remain possible and CSRF-token
protected. Make the equivalent distinction in docs/ja/concepts/studio.mdx at
lines 27-27, distinguishing remote access from local loopback connections.

In `@docs/cookbook/structured-outputs.mdx`:
- Around line 8-10: Revise the structured-output explanation around infer({
responseFormat }) to qualify the guarantee: supported successful responses are
constrained to the provided JSON Schema, but inference can still fail, return
empty content, or produce parsing/validation errors. Retain the existing runtime
validation and error-handling guidance, and remove claims that JSON.parse always
succeeds or that the result always contains the requested keys.

In `@docs/guides/cli/init.mdx`:
- Line 21: Restore the required npm-tab command form in the affected CLI
CodeGroups: use npm arkor followed by each existing subcommand instead of npx or
npm exec. Update docs/guides/cli/init.mdx:21-21, docs/ja/cli/auth.mdx:27-27,
89-89, and 137-137, docs/ja/cli/init.mdx:23-23, 135-135, 204-204, 230-230, and
256-256, docs/ja/guides/cli/init.mdx:21-21, docs/ja/quickstart.mdx:170-170, and
docs/quickstart.mdx:170-170; preserve each command’s existing subcommand and
options, including the CI, interactive, template-selection, default-options, and
OAuth login commands.

In `@docs/introduction.mdx`:
- Line 6: The lifecycle callback descriptions are ambiguous in both introduction
mirrors. Update docs/introduction.mdx at lines 6-6 to describe onLog,
onCheckpoint, onCompleted, and onFailed as user-implemented handlers that react
to lifecycle or training events; make the equivalent event-oriented wording
change in docs/ja/introduction.mdx at lines 6-6, replacing 「自分の関数として発火する」.

In `@docs/models.mdx`:
- Line 30: Update the roadmap wording in docs/models.mdx at lines 30-30 to use
“the Roadmap backlog” or equivalent, and update docs/ja/models.mdx at lines
30-30 to use the Japanese section name 「バックログ」 instead of “Backlog”.
- Line 6: Standardize the model identifier in the introductory prose and
createTrainer examples on both docs/models.mdx:6-6 and docs/ja/models.mdx:6-6.
Choose one canonical identifier, or explicitly explain the unsloth/ prefix when
retaining it in createTrainer, while preserving the English and Japanese
content.

In `@docs/studio/playground.mdx`:
- Line 37: Update the Playground guidance to explain that users should call the
infer function supplied inside onCheckpoint during checkpoint dispatch, rather
than presenting infer as a standalone SDK application API. Apply the equivalent
correction in docs/studio/playground.mdx:37-37 and
docs/ja/studio/playground.mdx:37-37, preserving each document’s language.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: fda7dc52-cb44-4dfc-ab6a-f43cb1046c86

📥 Commits

Reviewing files that changed from the base of the PR and between 71f5df1 and 5790d7a.

📒 Files selected for processing (39)

docs/cli/auth.mdx
docs/cli/init.mdx
docs/cli/overview.mdx
docs/concepts/lifecycle.mdx
docs/concepts/studio.mdx
docs/concepts/trainer.mdx
docs/cookbook/early-stopping.mdx
docs/cookbook/mid-run-eval.mdx
docs/cookbook/notifications.mdx
docs/cookbook/overview.mdx
docs/cookbook/structured-outputs.mdx
docs/guides/cli/auth.mdx
docs/guides/cli/dev.mdx
docs/guides/cli/init.mdx
docs/introduction.mdx
docs/ja/cli/auth.mdx
docs/ja/cli/init.mdx
docs/ja/cli/overview.mdx
docs/ja/concepts/lifecycle.mdx
docs/ja/concepts/studio.mdx
docs/ja/concepts/trainer.mdx
docs/ja/cookbook/early-stopping.mdx
docs/ja/cookbook/mid-run-eval.mdx
docs/ja/cookbook/notifications.mdx
docs/ja/cookbook/overview.mdx
docs/ja/cookbook/structured-outputs.mdx
docs/ja/guides/cli/auth.mdx
docs/ja/guides/cli/init.mdx
docs/ja/introduction.mdx
docs/ja/models.mdx
docs/ja/quickstart.mdx
docs/ja/roadmap.mdx
docs/ja/studio/overview.mdx
docs/ja/studio/playground.mdx
docs/models.mdx
docs/quickstart.mdx
docs/roadmap.mdx
docs/studio/overview.mdx
docs/studio/playground.mdx

📜 Review details

⏰ Context from checks skipped due to timeout. (75)

GitHub Check: cubic · AI code reviewer
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · windows-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · windows-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 22.22.0
GitHub Check: coverage · upload to Codecov
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · macos-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · windows-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=26.0.0 <27
GitHub Check: coverage · upload to Codecov

⚠️ CI failures not shown inline (2)

GitHub Actions: Running Copilot Code Review / copilot-pull-request-reviewer: Running Copilot Code Review

Conclusion: failure

View job details

##[group]Run set -euo pipefail
 �[36;1mset -euo pipefail�[0m
 �[36;1mecho "RUNNER_TEMP=$RUNNER_TEMP"�[0m
 �[36;1mfind "$RUNNER_TEMP" -maxdepth 1 -type f -name 'git-credentials-*.config' -print -delete�[0m
 �[36;1m�[0m
 �[36;1m# Generate a unique token and stop processing workflow commands to prevent the runtime from injecting commands�[0m
 �[36;1mSTOP_***REDACTED*** /proc/sys/kernel/random/uuid)�[0m
 �[36;1m�[0m
 �[36;1m# Use a trap to ensure we always resume command processing and check for�[0m
 �[36;1m# fallback error annotations, even if the runtime exits with a non-zero code�[0m
 �[36;1m# (which would otherwise cause set -e to abort the shell before we get here).�[0m
 �[36;1m# The trap preserves the original exit code.�[0m
 �[36;1mcopilot_cleanup() {�[0m
 �[36;1m  echo "::$STOP_***REDACTED***
 �[36;1m  FALLBACK_FILE="${RUNNER_TEMP}/copilot-fallback-error.txt"�[0m
 �[36;1m  if [ -f "$FALLBACK_FILE" ]; then�[0m
 �[36;1m    FALLBACK_MSG=$(head -c 500 "$FALLBACK_FILE" | tr -d '\n\r')�[0m
 �[36;1m    echo "::error title=Copilot Error::${FALLBACK_MSG}"�[0m

GitHub Actions: Running Copilot Code Review / 0_copilot-pull-request-reviewer.txt: Running Copilot Code Review

Conclusion: failure

View job details

##[group]Run set -euo pipefail
 �[36;1mset -euo pipefail�[0m
 �[36;1mecho "RUNNER_TEMP=$RUNNER_TEMP"�[0m
 �[36;1mfind "$RUNNER_TEMP" -maxdepth 1 -type f -name 'git-credentials-*.config' -print -delete�[0m
 �[36;1m�[0m
 �[36;1m# Generate a unique token and stop processing workflow commands to prevent the runtime from injecting commands�[0m
 �[36;1mSTOP_***REDACTED*** /proc/sys/kernel/random/uuid)�[0m
 �[36;1m�[0m
 �[36;1m# Use a trap to ensure we always resume command processing and check for�[0m
 �[36;1m# fallback error annotations, even if the runtime exits with a non-zero code�[0m
 �[36;1m# (which would otherwise cause set -e to abort the shell before we get here).�[0m
 �[36;1m# The trap preserves the original exit code.�[0m
 �[36;1mcopilot_cleanup() {�[0m
 �[36;1m  echo "::$STOP_***REDACTED***
 �[36;1m  FALLBACK_FILE="${RUNNER_TEMP}/copilot-fallback-error.txt"�[0m
 �[36;1m  if [ -f "$FALLBACK_FILE" ]; then�[0m
 �[36;1m    FALLBACK_MSG=$(head -c 500 "$FALLBACK_FILE" | tr -d '\n\r')�[0m
 �[36;1m    echo "::error title=Copilot Error::${FALLBACK_MSG}"�[0m

🧰 Additional context used

📓 Path-based instructions (2)

docs/**/*.mdx

📄 CodeRabbit inference engine (AGENTS.md)

docs/**/*.mdx: When editing English Mintlify documentation, update the corresponding Japanese mirror under docs/ja/ in the same change.
Cross-page Mintlify anchors must use the actual rendered heading IDs; verify them against the preview with curl before adding or editing links. Mintlify preserves /, =, and full-width parentheses but strips ASCII parentheses and backticks.

Files:

docs/guides/cli/dev.mdx
docs/guides/cli/init.mdx
docs/ja/studio/playground.mdx
docs/guides/cli/auth.mdx
docs/ja/guides/cli/init.mdx
docs/ja/concepts/trainer.mdx
docs/studio/playground.mdx
docs/concepts/lifecycle.mdx
docs/studio/overview.mdx
docs/ja/concepts/lifecycle.mdx
docs/cookbook/structured-outputs.mdx
docs/cli/auth.mdx
docs/concepts/trainer.mdx
docs/ja/cookbook/notifications.mdx
docs/ja/cookbook/overview.mdx
docs/ja/cli/auth.mdx
docs/ja/cookbook/structured-outputs.mdx
docs/cli/overview.mdx
docs/ja/cookbook/early-stopping.mdx
docs/cookbook/notifications.mdx
docs/cookbook/early-stopping.mdx
docs/ja/quickstart.mdx
docs/ja/guides/cli/auth.mdx
docs/ja/studio/overview.mdx
docs/ja/roadmap.mdx
docs/cookbook/overview.mdx
docs/ja/cli/overview.mdx
docs/concepts/studio.mdx
docs/cookbook/mid-run-eval.mdx
docs/quickstart.mdx
docs/cli/init.mdx
docs/roadmap.mdx
docs/ja/introduction.mdx
docs/ja/concepts/studio.mdx
docs/ja/cli/init.mdx
docs/models.mdx
docs/introduction.mdx
docs/ja/cookbook/mid-run-eval.mdx
docs/ja/models.mdx

docs/{cli,guides/cli}/**/*.mdx

📄 CodeRabbit inference engine (AGENTS.md)

CLI documentation CodeGroups must present unscripted subcommands as five tabs: pnpm, npm, yarn, yarn run, and bun; retain the npm arkor <subcommand> form and do not replace it with npx or npm exec.

Files:

docs/guides/cli/dev.mdx
docs/guides/cli/init.mdx
docs/guides/cli/auth.mdx
docs/cli/auth.mdx
docs/cli/overview.mdx
docs/cli/init.mdx

🪛 LanguageTool

docs/cookbook/mid-run-eval.mdx

[style] ~8-~8: Since ownership is already implied, this phrasing may be redundant.
Context: ...l into the partially trained model from your own code while the run is still going: each...

(PRP_OWN)

docs/introduction.mdx

[style] ~6-~6: Since ownership is already implied, this phrasing may be redundant.
Context: ...onCompleted, onFailed) that fire on your own functions, and a local Studio (`arkor d...

(PRP_OWN)

[style] ~16-~16: Since ownership is already implied, this phrasing may be redundant.
Context: ...redaction (PII redaction). - Train on your own dataset: a [Hugging Face dataset name o...

(PRP_OWN)

🔇 Additional comments (27)

docs/ja/concepts/lifecycle.mdx (1)

6-6: LGTM!

docs/ja/guides/cli/auth.mdx (1)

19-19: LGTM!

Also applies to: 47-47, 75-75

docs/introduction.mdx (1)

15-16: LGTM!

docs/ja/introduction.mdx (1)

16-16: LGTM!

docs/models.mdx (1)

21-22: LGTM!

docs/ja/models.mdx (1)

21-22: LGTM!

docs/ja/roadmap.mdx (1)

10-11: LGTM!

Also applies to: 81-81

docs/roadmap.mdx (1)

10-11: LGTM!

Also applies to: 81-81

docs/cookbook/early-stopping.mdx (1)

8-8: LGTM!

docs/cookbook/mid-run-eval.mdx (1)

8-8: LGTM!

Also applies to: 61-63

docs/cookbook/notifications.mdx (1)

8-8: LGTM!

docs/cookbook/overview.mdx (1)

6-6: LGTM!

Also applies to: 32-32

docs/ja/cookbook/early-stopping.mdx (1)

8-8: LGTM!

docs/ja/cookbook/mid-run-eval.mdx (1)

8-8: LGTM!

Also applies to: 63-63

docs/cli/init.mdx (1)

183-183: 🎯 Functional Correctness

Check these Mintlify heading anchors against the preview.

Use the rendered / and = fragments if Mintlify keeps them; otherwise update the cross-page links to the actual headings. Also confirm the create-arkor fragment on /quickstart.

Required preview checks:

/cli/init#ci-/-non-interactive-shells

/cli/init#claude-code-claudecode=1-strict-mode

/quickstart#running-under-claude-code-claudecode=1

docs/ja/cookbook/notifications.mdx (1)

8-8: LGTM!

docs/ja/cookbook/overview.mdx (1)

6-6: LGTM!

Also applies to: 32-32

docs/ja/cookbook/structured-outputs.mdx (1)

8-8: LGTM!

docs/ja/cli/init.mdx (1)

54-54: 🎯 Functional Correctness

Recheck the Mintlify preview anchors before merging.

Cross-page links must use the actual rendered heading IDs. Run the anchor checks against the preview origin for:

docs/ja/cli/init.mdx#L54: postinstall スクリプト（pnpm 11+）

docs/ja/cli/init.mdx#L183: Claude Code create-arkor target

docs/ja/cli/init.mdx#L287: postinstall スクリプト（pnpm 11+）

docs/ja/cli/overview.mdx#L35: CI / 非対話シェル

docs/ja/cli/overview.mdx#L36: Claude Code（CLAUDECODE=1）厳格モード

docs/ja/quickstart.mdx#L76: Claude Code create-arkor target

docs/quickstart.mdx#L76: English Claude Code strict-mode target

DOCS_PREVIEW_BASE_URL is required for this verification.

docs/concepts/lifecycle.mdx (1)

6-6: LGTM!

docs/concepts/studio.mdx (1)

60-60: LGTM!

docs/concepts/trainer.mdx (1)

142-142: LGTM!

docs/guides/cli/dev.mdx (1)

32-32: LGTM!

docs/ja/concepts/studio.mdx (1)

60-60: LGTM!

docs/ja/concepts/trainer.mdx (1)

142-142: LGTM!

docs/studio/overview.mdx (1)

71-71: LGTM!

docs/ja/studio/overview.mdx (1)

71-71: LGTM!

coderabbitai · 2026-07-24T08:45:09Z

+A fine-tuned model is supposed to emit a fixed shape (`triage` produces `{ category, urgency, summary, nextAction }`; `redaction` produces `{ redactedText, redactedCount, tags }`), but at a half-trained checkpoint the output drifts: extra prose, missing keys, the occasional unparseable blob.

 `infer({ responseFormat })` gets you a hard guarantee. The model is constrained at decode time to emit a string that matches the JSON Schema you hand it, so `JSON.parse` always succeeds and the resulting object always has the keys you asked for. That turns the mid-run check from "log a sample, eyeball it" into "extract typed fields, branch on them."


🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Qualify the structured-output guarantee.

The revised text removes the caveat while claiming that JSON Schema gives a “hard guarantee” and that JSON.parse “always succeeds.” The example itself still catches inference failures, empty content, and parsing errors; constrained decoding only applies to successful supported responses. State that the schema constrains valid model output, while retaining runtime validation and error handling.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/cookbook/structured-outputs.mdx` around lines 8 - 10, Revise the structured-output explanation around infer({ responseFormat }) to qualify the guarantee: supported successful responses are constrained to the provided JSON Schema, but inference can still fail, return empty content, or produce parsing/validation errors. Retain the existing runtime validation and error-handling guidance, and remove claims that JSON.parse always succeeds or that the result always contains the requested keys.

cubic-dev-ai

1 issue found across 39 files

Prompt for AI agents (unresolved issues)


Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="docs/cookbook/early-stopping.mdx">

<violation number="1" location="docs/cookbook/early-stopping.mdx:8">
P2: The English and Japanese cookbook pages are now out of sync: the Japanese page still publishes the introductory early-stopping claim that this change removes from English. Updating the corresponding Japanese lead in the same PR would preserve the required bilingual lockstep.</violation>
</file>

_{Reply with feedback, questions, or to request a fix.

Re-trigger cubic}

cubic-dev-ai · 2026-07-24T08:58:09Z

 # Early stopping on diverging loss

-If the loss starts climbing or NaNs out, the rest of the run is wasted compute. Arkor does not have built-in early stopping; it does have everything you need to bolt it on in a few lines of TypeScript.
+Arkor does not have built-in early stopping; it does have everything you need to bolt it on in a few lines of TypeScript.


P2: The English and Japanese cookbook pages are now out of sync: the Japanese page still publishes the introductory early-stopping claim that this change removes from English. Updating the corresponding Japanese lead in the same PR would preserve the required bilingual lockstep.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At docs/cookbook/early-stopping.mdx, line 8: <comment>The English and Japanese cookbook pages are now out of sync: the Japanese page still publishes the introductory early-stopping claim that this change removes from English. Updating the corresponding Japanese lead in the same PR would preserve the required bilingual lockstep.</comment> <file context> @@ -5,7 +5,7 @@ description: "Abort a run automatically when the loss curve goes the wrong way, # Early stopping on diverging loss -If the loss starts climbing or NaNs out, the rest of the run is wasted compute. Arkor does not have built-in early stopping; it does have everything you need to bolt it on in a few lines of TypeScript. +Arkor does not have built-in early stopping; it does have everything you need to bolt it on in a few lines of TypeScript. This recipe uses three primitives together: </file context>

Restores the two documented conventions this PR wrongly 'fixed': the npm CodeGroup tabs return to the 'npm arkor <subcommand>' form (AGENTS .md documents the 5-tab presentation convention; the tabs are not meant to be copy-pasteable without a project), and anchor fragments return to Mintlify's actual heading ids, which preserve '=', '/', and full-width parens (AGENTS.md documents Mintlify's mixed slug behavior; the GitHub-style ids this PR substituted do not exist in the rendered docs). Also removes three em dashes the new sentences introduced (CI bans them). Substantive review fixes, EN and JA in lockstep: the Playground system-prompt tip routes through onCheckpoint's infer (infer is not a standalone API); concepts/studio no longer calls loopback 'no inbound connection' (local requests still arrive, gated by the token); the structured-output guarantee gains its maxTokens-truncation caveat; the introduction says callbacks 'invoke the functions you register'; the models intro drops the wrong only-registered-models-chat claim; the roadmap card no longer calls the model field single-valued; the early-stopping lead is restored in English to match the Japanese mirror; and mid-run-eval scopes the base-comparison limitation to the callback's infer, noting CloudApiClient.chat({ baseModel }) as the direct path.

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ed8640d74b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-07-25T10:03:04Z

 ## Variations

-**Compare against the base model on the same prompt.** Studio's Playground already has a Base / Adapter mode toggle, but you can do the same thing from `onCheckpoint` to score automatically rather than eyeballing it.
+**Forward each checkpoint's sample to a review queue.** The callback's `infer` is permanently bound to the just-saved checkpoint and `InferArgs` has no base-model override, so this callback alone cannot produce a base-vs-adapter comparison (issue a separate base-model request via `CloudApiClient.chat({ baseModel })`, or use Studio's Playground Base / Adapter toggle manually). What the callback can automate by itself is capturing one sample per checkpoint for later review:


Show the actual CloudApiClient.chat input shape

For readers implementing the suggested automated base-vs-adapter comparison, CloudApiClient.chat({ baseModel }) cannot work: the public method requires scope plus a body containing both messages and baseModel (packages/arkor/src/core/client.ts:362-379). Show a valid call such as client.chat({ scope, body: { baseModel, messages } }); the same invalid shorthand also appears in the Japanese mirror.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-07-25T10:03:04Z

 - **Intermediate checkpoints.** The dropdown only lists completed jobs and only sends `kind: "final"`. To run inference during a run, use the SDK's [`onCheckpoint({ infer })`](/sdk/callbacks): the `infer` it gives you is bound to the checkpoint that just landed.
 - **`temperature` / `topP` / `maxTokens`.** The HTTP shape ([`InferArgs`](/sdk/infer)) accepts these, but the Playground does not surface input fields for them. Studio sends only `messages` and the mode-specific fields. To experiment with these knobs, call the SDK's `infer` directly with the values you want.
- **System prompts.** The input field only adds a `user` role. There is no field for `system`. Edit `src/arkor/trainer.ts` (or call `infer` programmatically) if you want to test with a system prompt.
+- **System prompts.** The input field only adds a `user` role. There is no field for `system`, and the Playground builds its request solely from what you type on this page. To test with a system prompt, pass a `system` message to the [`infer`](/sdk/infer) handed to [`onCheckpoint`](/sdk/callbacks) during a run.


Describe the assistant turns included in Playground requests

On the second and later turns, the Playground does not build the request solely from text the user typed: Playground.send() passes messages: [...messages, userMsg], and messages includes the assistant responses accumulated from earlier streams (packages/studio-app/src/pages/Playground.tsx:166-205). Saying otherwise misdocuments its multi-turn context; describe it as the conversation accumulated on this page instead. The Japanese mirror repeats the same claim.

Useful? React with 👍 / 👎.

coderabbitai

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/cookbook/mid-run-eval.mdx`:
- Line 63: Update the CloudApiClient.chat guidance in
docs/cookbook/mid-run-eval.mdx at lines 63-63 to show the required scope
property and nest baseModel under body, or reference a valid public helper.
Apply the same corrected request shape in docs/ja/cookbook/mid-run-eval.mdx at
lines 63-63, preserving the surrounding checkpoint-evaluation explanation.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 926549f8-9c47-45a0-9f45-2220c8b90344

📥 Commits

Reviewing files that changed from the base of the PR and between 5790d7a and ed8640d.

📒 Files selected for processing (17)

docs/concepts/studio.mdx
docs/cookbook/early-stopping.mdx
docs/cookbook/mid-run-eval.mdx
docs/cookbook/overview.mdx
docs/cookbook/structured-outputs.mdx
docs/introduction.mdx
docs/ja/concepts/studio.mdx
docs/ja/cookbook/mid-run-eval.mdx
docs/ja/cookbook/structured-outputs.mdx
docs/ja/introduction.mdx
docs/ja/models.mdx
docs/ja/roadmap.mdx
docs/ja/studio/playground.mdx
docs/models.mdx
docs/roadmap.mdx
docs/studio/overview.mdx
docs/studio/playground.mdx

📜 Review details

⏰ Context from checks skipped due to timeout. (279)

GitHub Check: install · pnpm-9 · macos-latest · node >=24.12.0 <25
GitHub Check: install · yarn · macos-latest · node 24.0.0
GitHub Check: install · npm · windows-latest · node 26.0.0
GitHub Check: install · yarn-berry · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-9 · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-9 · macos-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · windows-latest · node >=26.0.0 <27
GitHub Check: install · bun · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-11 · windows-latest · node 24.12.0
GitHub Check: install · yarn · windows-latest · node >=26.0.0 <27
GitHub Check: install · yarn-berry · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-10 · macos-latest · node 22.22.0
GitHub Check: install · npm · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-11 · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-11 · windows-latest · node 26.0.0
GitHub Check: install · yarn · windows-latest · node 26.0.0
GitHub Check: install · pnpm-9 · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-10 · windows-latest · node >=24.12.0 <25
GitHub Check: install · bun · windows-latest · node 26.0.0
GitHub Check: install · bun · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-9 · windows-latest · node 26.0.0
GitHub Check: install · yarn-berry · windows-latest · node 26.0.0
GitHub Check: install · npm · windows-latest · node >=24.12.0 <25
GitHub Check: install · bun · windows-latest · node 24.12.0
GitHub Check: install · pnpm-10 · windows-latest · node 24.12.0
GitHub Check: install · pnpm-10 · windows-latest · node 26.0.0
GitHub Check: install · pnpm-10 · windows-latest · node 24.0.0
GitHub Check: install · pnpm-11 · windows-latest · node 24.0.0
GitHub Check: install · yarn · windows-latest · node 24.0.0
GitHub Check: install · yarn · windows-latest · node >=24.12.0 <25
GitHub Check: install · npm · windows-latest · node 24.0.0
GitHub Check: install · bun · windows-latest · node 24.0.0
GitHub Check: install · pnpm-11 · windows-latest · node >=24.12.0 <25
GitHub Check: install · yarn · windows-latest · node 24.12.0
GitHub Check: install · npm · windows-latest · node 24.12.0
GitHub Check: install · yarn-berry · windows-latest · node 24.12.0
GitHub Check: install · pnpm-9 · windows-latest · node 24.12.0
GitHub Check: install · yarn-berry · windows-latest · node 24.0.0
GitHub Check: install · pnpm-9 · windows-latest · node 24.0.0
GitHub Check: install · bun · windows-latest · node 22.22.0
GitHub Check: install · yarn · windows-latest · node 22.22.0
GitHub Check: install · yarn · windows-latest · node >=22.22.0 <23
GitHub Check: install · npm · windows-latest · node >=22.22.0 <23
GitHub Check: install · bun · windows-latest · node >=22.22.0 <23
GitHub Check: install · bun · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-11 · windows-latest · node 22.22.0
GitHub Check: install · yarn-berry · windows-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · windows-latest · node 22.22.0
GitHub Check: install · bun · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-9 · windows-latest · node 22.22.0
GitHub Check: install · yarn-berry · windows-latest · node 22.22.0
GitHub Check: install · yarn · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-9 · windows-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · yarn-berry · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · npm · windows-latest · node 22.22.0
GitHub Check: install · pnpm-10 · windows-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · windows-latest · node >=22.22.0 <23
GitHub Check: install · yarn · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-10 · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · npm · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-9 · ubuntu-latest · node 26.0.0
GitHub Check: install · yarn-berry · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · yarn · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · yarn · ubuntu-latest · node 24.0.0
GitHub Check: install · yarn-berry · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · bun · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-11 · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-11 · ubuntu-latest · node 26.0.0
GitHub Check: install · npm · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-10 · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node 24.0.0
GitHub Check: install · npm · ubuntu-latest · node 24.0.0
GitHub Check: install · bun · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · ubuntu-latest · node 24.0.0
GitHub Check: install · npm · ubuntu-latest · node 24.12.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · bun · ubuntu-latest · node 24.12.0
GitHub Check: install · npm · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn-berry · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn · ubuntu-latest · node 24.12.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn-berry · ubuntu-latest · node 22.22.0
GitHub Check: install · yarn-berry · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node 24.12.0
GitHub Check: install · npm · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · bun · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · yarn-berry · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · bun · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · yarn · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · ubuntu-latest · node 22.22.0
GitHub Check: install · yarn · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node 22.22.0
GitHub Check: install · yarn-berry · macos-latest · node 24.12.0
GitHub Check: install · pnpm-10 · macos-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-9 · macos-latest · node 24.0.0
GitHub Check: install · bun · macos-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · macos-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · macos-latest · node 22.22.0
GitHub Check: install · yarn · macos-latest · node 22.22.0
GitHub Check: install · npm · macos-latest · node >=22.22.0 <23
GitHub Check: install · yarn · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-10 · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-11 · windows-latest · node >=26.0.0 <27
GitHub Check: install · bun · windows-latest · node 26.0.0
GitHub Check: install · pnpm-9 · macos-latest · node >=22.22.0 <23
GitHub Check: install · yarn-berry · windows-latest · node 26.0.0
GitHub Check: install · yarn-berry · windows-latest · node 24.12.0
GitHub Check: install · pnpm-11 · windows-latest · node 24.12.0
GitHub Check: install · npm · windows-latest · node >=26.0.0 <27
GitHub Check: install · bun · macos-latest · node 22.22.0
GitHub Check: install · yarn-berry · macos-latest · node 22.22.0
GitHub Check: install · pnpm-9 · windows-latest · node >=26.0.0 <27
GitHub Check: install · yarn-berry · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-11 · macos-latest · node >=22.22.0 <23
GitHub Check: install · bun · windows-latest · node >=26.0.0 <27
GitHub Check: install · npm · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-10 · windows-latest · node 26.0.0
GitHub Check: install · yarn-berry · windows-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-10 · windows-latest · node >=26.0.0 <27
GitHub Check: install · npm · windows-latest · node 24.0.0
GitHub Check: install · bun · windows-latest · node 24.12.0
GitHub Check: install · npm · windows-latest · node 24.12.0
GitHub Check: install · pnpm-11 · windows-latest · node 24.0.0
GitHub Check: install · npm · windows-latest · node 26.0.0
GitHub Check: install · yarn · windows-latest · node 26.0.0
GitHub Check: install · yarn · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-10 · windows-latest · node 24.12.0
GitHub Check: install · pnpm-9 · windows-latest · node 26.0.0
GitHub Check: install · pnpm-11 · windows-latest · node 26.0.0
GitHub Check: install · bun · windows-latest · node 24.0.0
GitHub Check: install · bun · windows-latest · node >=24.12.0 <25
GitHub Check: install · yarn-berry · windows-latest · node 24.0.0
GitHub Check: install · pnpm-9 · windows-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-11 · windows-latest · node >=24.12.0 <25
GitHub Check: install · yarn · windows-latest · node 24.12.0
GitHub Check: install · pnpm-9 · windows-latest · node 24.12.0
GitHub Check: install · pnpm-10 · windows-latest · node 24.0.0
GitHub Check: install · yarn · windows-latest · node 24.0.0
GitHub Check: install · pnpm-9 · windows-latest · node 24.0.0
GitHub Check: install · pnpm-11 · windows-latest · node >=22.22.0 <23
GitHub Check: install · bun · windows-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-10 · windows-latest · node >=22.22.0 <23
GitHub Check: install · yarn-berry · windows-latest · node >=22.22.0 <23
GitHub Check: install · npm · ubuntu-latest · node 26.0.0
GitHub Check: install · npm · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-10 · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · yarn-berry · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-9 · windows-latest · node 22.22.0
GitHub Check: install · yarn · ubuntu-latest · node 26.0.0
GitHub Check: install · yarn · windows-latest · node 22.22.0
GitHub Check: install · pnpm-9 · windows-latest · node >=22.22.0 <23
GitHub Check: install · npm · windows-latest · node >=22.22.0 <23
GitHub Check: install · yarn · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · yarn · windows-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · npm · windows-latest · node 22.22.0
GitHub Check: install · pnpm-10 · windows-latest · node 22.22.0
GitHub Check: install · pnpm-11 · windows-latest · node 22.22.0
GitHub Check: install · yarn-berry · windows-latest · node 22.22.0
GitHub Check: install · bun · windows-latest · node 22.22.0
GitHub Check: install · yarn-berry · ubuntu-latest · node >=26.0.0 <27
GitHub Check: install · pnpm-9 · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · yarn-berry · ubuntu-latest · node 24.12.0
GitHub Check: install · bun · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node 26.0.0
GitHub Check: install · yarn · ubuntu-latest · node 24.12.0
GitHub Check: install · bun · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-9 · ubuntu-latest · node 26.0.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn-berry · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · npm · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · yarn-berry · ubuntu-latest · node 24.0.0
GitHub Check: install · bun · ubuntu-latest · node 24.12.0
GitHub Check: install · bun · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn · ubuntu-latest · node >=24.12.0 <25
GitHub Check: install · pnpm-9 · ubuntu-latest · node 24.0.0
GitHub Check: install · npm · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node 24.12.0
GitHub Check: install · yarn · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-9 · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · yarn · ubuntu-latest · node 22.22.0
GitHub Check: install · npm · ubuntu-latest · node 22.22.0
GitHub Check: install · npm · ubuntu-latest · node 24.0.0
GitHub Check: install · yarn-berry · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · yarn-berry · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-9 · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-11 · ubuntu-latest · node 24.0.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node 22.22.0
GitHub Check: install · pnpm-10 · ubuntu-latest · node >=22.22.0 <23
GitHub Check: install · pnpm-11 · ubuntu-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · macos-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · windows-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · windows-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.3.0 <24.5.0
GitHub Check: coverage · upload to Codecov
GitHub Check: Analyze (javascript-typescript)
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=26.0.0 <27
GitHub Check: typecheck · lint · test · build · windows-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · macos-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · macos-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · macos-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.10.0 <24.12.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 24.12.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.8.0 <24.10.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 26.0.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.0.0 <24.1.0
GitHub Check: typecheck · lint · test · build · windows-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node 22.22.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.1.0 <24.3.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=24.5.0 <24.8.0
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.12.0 <25
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=24.3.0 <24.5.0
GitHub Check: typecheck · lint · test · build · windows-latest · node >=22.22.0 <23
GitHub Check: typecheck · lint · test · build · ubuntu-latest · node >=22.22.0 <23
GitHub Check: Analyze (javascript-typescript)
GitHub Check: coverage · upload to Codecov

⚠️ CI failures not shown inline (2)

GitHub Actions: Running Copilot Code Review / copilot-pull-request-reviewer: Running Copilot Code Review

Conclusion: failure

View job details

##[group]Run set -euo pipefail
 �[36;1mset -euo pipefail�[0m
 �[36;1mecho "RUNNER_TEMP=$RUNNER_TEMP"�[0m
 �[36;1mfind "$RUNNER_TEMP" -maxdepth 1 -type f -name 'git-credentials-*.config' -print -delete�[0m
 �[36;1m�[0m
 �[36;1m# Generate a unique token and stop processing workflow commands to prevent the runtime from injecting commands�[0m
 �[36;1mSTOP_***REDACTED*** /proc/sys/kernel/random/uuid)�[0m
 �[36;1m�[0m
 �[36;1m# Use a trap to ensure we always resume command processing and check for�[0m
 �[36;1m# fallback error annotations, even if the runtime exits with a non-zero code�[0m
 �[36;1m# (which would otherwise cause set -e to abort the shell before we get here).�[0m
 �[36;1m# The trap preserves the original exit code.�[0m
 �[36;1mcopilot_cleanup() {�[0m
 �[36;1m  echo "::$STOP_***REDACTED***
 �[36;1m  FALLBACK_FILE="${RUNNER_TEMP}/copilot-fallback-error.txt"�[0m
 �[36;1m  if [ -f "$FALLBACK_FILE" ]; then�[0m
 �[36;1m    FALLBACK_MSG=$(head -c 500 "$FALLBACK_FILE" | tr -d '\n\r')�[0m
 �[36;1m    echo "::error title=Copilot Error::${FALLBACK_MSG}"�[0m

GitHub Actions: Running Copilot Code Review / 0_copilot-pull-request-reviewer.txt: Running Copilot Code Review

Conclusion: failure

View job details

##[group]Run set -euo pipefail
 �[36;1mset -euo pipefail�[0m
 �[36;1mecho "RUNNER_TEMP=$RUNNER_TEMP"�[0m
 �[36;1mfind "$RUNNER_TEMP" -maxdepth 1 -type f -name 'git-credentials-*.config' -print -delete�[0m
 �[36;1m�[0m
 �[36;1m# Generate a unique token and stop processing workflow commands to prevent the runtime from injecting commands�[0m
 �[36;1mSTOP_***REDACTED*** /proc/sys/kernel/random/uuid)�[0m
 �[36;1m�[0m
 �[36;1m# Use a trap to ensure we always resume command processing and check for�[0m
 �[36;1m# fallback error annotations, even if the runtime exits with a non-zero code�[0m
 �[36;1m# (which would otherwise cause set -e to abort the shell before we get here).�[0m
 �[36;1m# The trap preserves the original exit code.�[0m
 �[36;1mcopilot_cleanup() {�[0m
 �[36;1m  echo "::$STOP_***REDACTED***
 �[36;1m  FALLBACK_FILE="${RUNNER_TEMP}/copilot-fallback-error.txt"�[0m
 �[36;1m  if [ -f "$FALLBACK_FILE" ]; then�[0m
 �[36;1m    FALLBACK_MSG=$(head -c 500 "$FALLBACK_FILE" | tr -d '\n\r')�[0m
 �[36;1m    echo "::error title=Copilot Error::${FALLBACK_MSG}"�[0m

🧰 Additional context used

📓 Path-based instructions (2)

docs/**/*.mdx

📄 CodeRabbit inference engine (AGENTS.md)

Keep English and Japanese documentation paired: changes under docs/ must also update the corresponding files under docs/ja/. Verify Mintlify-generated anchors before adding cross-page links; preserve /, =, and full-width parentheses while accounting for stripped ASCII punctuation and backticks.

Files:

docs/ja/studio/playground.mdx
docs/studio/overview.mdx
docs/concepts/studio.mdx
docs/studio/playground.mdx
docs/introduction.mdx
docs/ja/cookbook/structured-outputs.mdx
docs/cookbook/structured-outputs.mdx
docs/cookbook/early-stopping.mdx
docs/models.mdx
docs/cookbook/mid-run-eval.mdx
docs/ja/roadmap.mdx
docs/cookbook/overview.mdx
docs/ja/introduction.mdx
docs/roadmap.mdx
docs/ja/cookbook/mid-run-eval.mdx
docs/ja/concepts/studio.mdx
docs/ja/models.mdx

**/{*.md,*.mdx,*.yaml,*.yml}

📄 CodeRabbit inference engine (AGENTS.md)

Do not format Markdown, MDX, YAML, or YML files with oxfmt; these are excluded because documentation anchors and deliberate YAML layout must remain hand-managed.

Files:

docs/ja/studio/playground.mdx
docs/studio/overview.mdx
docs/concepts/studio.mdx
docs/studio/playground.mdx
docs/introduction.mdx
docs/ja/cookbook/structured-outputs.mdx
docs/cookbook/structured-outputs.mdx
docs/cookbook/early-stopping.mdx
docs/models.mdx
docs/cookbook/mid-run-eval.mdx
docs/ja/roadmap.mdx
docs/cookbook/overview.mdx
docs/ja/introduction.mdx
docs/roadmap.mdx
docs/ja/cookbook/mid-run-eval.mdx
docs/ja/concepts/studio.mdx
docs/ja/models.mdx

🔇 Additional comments (15)

docs/models.mdx (2)

6-6: Keep the hosted model identifier consistent with the implementation.

packages/studio-app/src/lib/baseModels.ts lists unsloth/gemma-4-E4B-it, while this paragraph says hosted inference serves gemma-4-E4B-it. Use the canonical identifier consistently, or explicitly document that the public API intentionally strips the unsloth/ prefix.

30-30: LGTM!

docs/ja/models.mdx (2)

6-6: ホスト型モデルの識別子を実装と一致させてください。

packages/studio-app/src/lib/baseModels.ts は unsloth/gemma-4-E4B-it を定義していますが、この説明では gemma-4-E4B-it と記載されています。正規の識別子に統一するか、公開 API が unsloth/ 接頭辞を意図的に省略することを明記してください。

30-30: LGTM!

docs/concepts/studio.mdx (1)

27-27: LGTM!

Also applies to: 60-60

docs/ja/concepts/studio.mdx (1)

27-27: LGTM!

docs/ja/studio/playground.mdx (1)

37-37: LGTM!

docs/studio/overview.mdx (1)

71-71: LGTM!

docs/studio/playground.mdx (1)

37-37: LGTM!

docs/ja/roadmap.mdx (1)

10-15: LGTM!

Also applies to: 81-81

docs/roadmap.mdx (1)

10-15: LGTM!

Also applies to: 81-81

docs/cookbook/early-stopping.mdx (1)

8-8: LGTM!

docs/cookbook/mid-run-eval.mdx (1)

8-8: LGTM!

docs/ja/cookbook/mid-run-eval.mdx (1)

8-8: LGTM!

docs/cookbook/overview.mdx (1)

32-32: LGTM!

coderabbitai · 2026-07-25T10:04:33Z

 ## Variations

-**Compare against the base model on the same prompt.** Studio's Playground already has a Base / Adapter mode toggle, but you can do the same thing from `onCheckpoint` to score automatically rather than eyeballing it.
+**Forward each checkpoint's sample to a review queue.** The callback's `infer` is permanently bound to the just-saved checkpoint and `InferArgs` has no base-model override, so this callback alone cannot produce a base-vs-adapter comparison (issue a separate base-model request via `CloudApiClient.chat({ baseModel })`, or use Studio's Playground Base / Adapter toggle manually). What the callback can automate by itself is capturing one sample per checkpoint for later review:


🎯 Functional Correctness | 🟠 Major | ⚡ Quick win

Use the actual CloudApiClient.chat request shape.

The documented CloudApiClient.chat({ baseModel }) call omits the required scope and places baseModel at the wrong level; the SDK expects it under body.

docs/cookbook/mid-run-eval.mdx#L63-L63: show a valid chat({ scope, body: { baseModel, ... } }) request or point to a valid public helper.

docs/ja/cookbook/mid-run-eval.mdx#L63-L63: apply the same corrected request shape in the Japanese explanation.

📍 Affects 2 files

docs/cookbook/mid-run-eval.mdx#L63-L63 (this comment)

docs/ja/cookbook/mid-run-eval.mdx#L63-L63

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/cookbook/mid-run-eval.mdx` at line 63, Update the CloudApiClient.chat guidance in docs/cookbook/mid-run-eval.mdx at lines 63-63 to show the required scope property and nest baseModel under body, or reference a valid public helper. Apply the same corrected request shape in docs/ja/cookbook/mid-run-eval.mdx at lines 63-63, preserving the surrounding checkpoint-evaluation explanation.

Nicolas0315 · 2026-08-04T06:17:56Z

The windows-latest · node >=24.0.0 <24.1.0 failure on this PR is an unrelated flake: the /api/manifest tests run a real esbuild bundle under vitest's default 5 s timeout, and slow Windows shards occasionally exceed it (Error: Test timed out in 5000ms in the job log — the No credentials on file line above it is console noise from another test). Root cause + fix in #213; a re-run should go green.

…imeout Each test in the /api/manifest describe runs a real esbuild bundle via readManifestSummary but sat under vitest's default 5s testTimeout. Fast machines finish the block in ~250ms; slow windows-latest CI shards have blown the budget (observed on PR #203, a docs-only change). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

soleil-colza added 3 commits July 23, 2026 22:09

Copilot AI review requested due to automatic review settings July 24, 2026 08:38

Copilot started reviewing on behalf of soleil-colza July 24, 2026 08:38 View session

Copilot AI reviewed Jul 24, 2026

chatgpt-codex-connector Bot reviewed Jul 24, 2026

View reviewed changes

greptile-apps Bot reviewed Jul 24, 2026

View reviewed changes

Comment thread docs/cli/auth.mdx Outdated

Comment thread docs/concepts/studio.mdx Outdated

coderabbitai Bot requested changes Jul 24, 2026

View reviewed changes

cubic-dev-ai Bot reviewed Jul 24, 2026

View reviewed changes

soleil-colza added 2 commits July 25, 2026 02:58

Merge remote-tracking branch 'origin/main' into eng-955

ed8640d

Copilot AI review requested due to automatic review settings July 25, 2026 09:58

Copilot started reviewing on behalf of soleil-colza July 25, 2026 09:59 View session

Copilot AI reviewed Jul 25, 2026

chatgpt-codex-connector Bot reviewed Jul 25, 2026

View reviewed changes

coderabbitai Bot requested changes Jul 25, 2026

View reviewed changes

coderabbitai Bot mentioned this pull request Aug 3, 2026

feat(arkor): type the createTrainer model field so unsupported base models fail at compile time #211

Merged

Nicolas0315 mentioned this pull request Aug 4, 2026

fix(arkor): give the esbuild-backed /api/manifest tests a realistic timeout #213

Merged

coderabbitai Bot mentioned this pull request Aug 4, 2026

fix(dev): fall back to next free port when --port is not explicit #212

Merged

coderabbitai Bot mentioned this pull request Aug 16, 2026

feat: local MLX training and inference behind arkor start/dev --local (ENG-1065) #228

Open

		\| `CI=1` (or any non-TTY stdout) \| Force non-interactive mode. `arkor init` skips its prompts and uses the values you passed via flags (or built-in defaults under `--yes`); git init runs without asking when `--git` or `--yes` is set, and is skipped silently otherwise. See [`arkor init` § CI / non-interactive shells](/cli/init#ci--non-interactive-shells). \|
		\| `CLAUDECODE=1` \| Strict mode for AI agents that spawn the CLI without being able to answer interactive prompts. `arkor init` (and `create-arkor`) refuse to run unless every curated flag is present, printing a multi-line stderr block with the suggested re-invocation; pass `-y`/`--yes` to opt back into "accept all defaults". See [`arkor init` § Claude Code (`CLAUDECODE=1`) strict mode](/cli/init#claude-code-claudecode1-strict-mode). \|

		A fine-tuned model is supposed to emit a fixed shape (`triage` produces `{ category, urgency, summary, nextAction }`; `redaction` produces `{ redactedText, redactedCount, tags }`), but at a half-trained checkpoint the output drifts: extra prose, missing keys, the occasional unparseable blob.

		`infer({ responseFormat })` gets you a hard guarantee. The model is constrained at decode time to emit a string that matches the JSON Schema you hand it, so `JSON.parse` always succeeds and the resulting object always has the keys you asked for. That turns the mid-run check from "log a sample, eyeball it" into "extract typed fields, branch on them."

Uh oh!

Conversation

soleil-colza commented Jul 24, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Summary by cubic

Summary by CodeRabbit

Uh oh!

qodo-code-review Bot commented Jul 24, 2026

Uh oh!

coderabbitai Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

drift-check Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Code Review Bot

Uh oh!

github-code-quality Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Code Coverage Overview

TypeScript / code-coverage/arkor

TypeScript / code-coverage/create-arkor

TypeScript / code-coverage/cli-internal

TypeScript / code-coverage/studio-app

Uh oh!

codecov Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cubic-dev-ai Bot Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

soleil-colza commented Jul 24, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Jul 24, 2026 •

edited

Loading

drift-check Bot commented Jul 24, 2026 •

edited

Loading

github-code-quality Bot commented Jul 24, 2026 •

edited

Loading

codecov Bot commented Jul 24, 2026 •

edited

Loading

greptile-apps Bot commented Jul 24, 2026 •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading