Repository navigation
docs: walk through app design flow and clarify runtime boundaries - #143
Merged
Merged
Conversation
This was referenced Oct 2, 2026
Merged
Merged
ymote
commented
Oct 2, 2026
ymote
left a comment
Contributor
Author
There was a problem hiding this comment.
Review: needs changes before merge
Accuracy: B. Writing: C.
Checked against main caa5d36 (the PR base) and the current lock pin cb66de07 (makepad 1f3b1ded, Octoscript dbd48cfb). Correct: tools/octo binary lookup, run/--detach, shot waiting for two identical frames, check and stamp handling, the flow.py artboard, scene count, locales and stages, card-host registering no services, offered_tools defaults, and the sys.chat caveat (Octoscript #53 is after dbd48cfb). check-links.py and test_octo_run.py pass.
The README and AGENTS.md corrections (--detach, check forwarding, where L0 lives) are good as they are.
Accuracy
- Cross-repo links. §2 (
:47-50) and §7 link to App Hubdocs/CODE-WALKTHROUGH.mdand OctoSensedocs/architecture-walkthrough.md, which only exist in App Hub #55 and OctoSense #275.check-links.pyskips external URLs, so CI won't catch this. Merge after those. - Parts of the flow are missing.
kits/sketchisn't mentioned.- Neither is
flows/image-lib. - Neither is
beauty-pipeline.sh, whichimage-to-appcard-flow.shactually calls. - The hand-off stops at
check.flows/README.mddefines stamp → check → card-host → screenshot (human) → restamp → sign (human) → submit (human). Name the human steps. - §8 should give the default
--stages(intake,semantic,compile,bundle,service-test).
- Snapshots.
:5("Design Flowcaa5d364and App Hub41bc959") and the full lock hash at:270will go stale; Octoscript-Makepadmainis already 10 commits pastcb66de07. "At the reviewed shell snapshot" (:194) and "At this snapshot" (:211) don't say which one. Link the pin files. - §1 heading. "Three kinds of source, two kinds of agent" doesn't match its table, which has seven mixed rows including the CLI and octos.
- Exit status. In the README, "forwards exit status" has one exception: if
hub stampfails,cmd_checkreturns that status and never runshub check.
Writing
The facts are dense and correct, but the voice is a reviewer's, not a manual's.
| Where | Current | Suggested |
|---|---|---|
:3 |
For a junior Rust developer, begin with the executable path… | This repository is a Python harness. The Rust runtime lives in App Hub and the agent kernel in octos. |
:5-7 |
This review follows… unverified in this review… | Delete. |
:306-312 |
Review verification (2026-10-01)… | Delete; put it in the PR description. |
:90 |
A file existing is not a visual review: open the capture. | Open the PNG and look at it. |
:95-96 |
…so the gate output is forwarded, not necessarily the CLI's only output | check prints the hub check output and exit status, plus its own progress lines. |
:136-137 |
…the walkthrough does not claim it is publication-ready | The untouched template fails the full gate until the listing and screenshots are done. |
:264-265 |
…a repository code review does not need to invent app screenshots | Delete. |
:230 |
…a peer identity, conversation context, running future and worker thread are different objects | One peer can hold several conversation contexts over its lifetime; it is not a thread or a task. |
Also:
:192shows<app>as raw HTML entities. Use`Ask <app>`.- The
AGENTS.mdphrase "without pretending to complete an app-authoring flow" has the same reviewer voice. - README "Code walkthrough for Rust beginners" → "Code walkthrough".
- Two over-long lines:
README.md("…Octoscript L0 (page.card). Top-level…") and §7.
zh-CN
README.zh-CN.mdadds the "main.splashvs L0 parsing path" sentence in two places; the English has it once.- "从 Python CLI…追踪到 App Hub 与 OctoSense Shell" reads like a translation. Try "沿着 Python CLI、脚本模板…一路讲到 App Hub 与 OctoSense Shell".
- 封装层进度和 listing 占位符提示 → 清单(listing)占位提示.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add a walkthrough from the Python CLI and script template through runtime preparation, bundle admission and publication. Correct
run --detachguidance andcheckbehavior in the English/Chinese README and contributor instructions.The guide maps the sketch kit, image library, pipeline scripts, default stages, fail-fast receipts and human screenshot/restamp/sign/submit steps. It distinguishes the system peer session from human conversation contexts, qualifies Unix workspace tools, and states that the current authoring pin lacks
sys.chat/ChatEntry. Runtime versions are read from lock files; companion guide links resolve before merge.Writing revision for junior Rust readers: Define peer, slug, session, shared context, event sink and blackboard within one request. Trace the five default image stages through concrete inputs and outputs before optional stages.
Validation: six CLI tests and all five example plans pass; the repository link checker reports zero failures; relative file links, remote companion links and whitespace pass. Native UI, provider conversations and publishing credentials were not exercised.
Writing assessment after parallel revision and a separate review: 4.5/5 (A−), considering beginner orientation, terminology, worked examples, organization and consistency. This supersedes the earlier combined technical/writing score where present; it is our assessment, not an external approval.
Companion PRs: