Conversation
…r two The generated block on reference-js was 47 lines against 1599 for Go and 1554 for Python. Most of that gap is honest - the Node client has no generated facade to document - but part of it was the generator walking classes and stopping there, and part was it emitting a flat page where the sibling pages emit a grouped index a reader can scan. So it now lifts constructors, the public fields a constructor sets, exported functions and exported constants, and groups them the way sync-api-go.py and sync-api-python.py group theirs: the workbench first, then errors under one heading, then module functions, then constants, each group listed in the index at the top. Methods are qualified with their class, as Python's are, so a heading read on its own says which type it belongs to. A constant gets its own heading and its declaration in a fence, so PROTOCOL reads as the wire version the module exports rather than as a 1 in a bullet list. Private members stay out. `_onData` and `_fail` are how the client happens to work today, and writing them down would promise a caller something the next commit is free to take away. The preamble no longer says the page is generated from the JSDoc, because the generated block's first line now says so; the room it freed goes to the thing the generator cannot say, which is why there is no facade here and where to find the verb names instead. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The JSDoc improvements landed in the application repository, so the page is regenerated against them rather than against the branch it was written on. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ne agent MeshBench maintains three agent skills beside its code and mirrors them into two standalone repositories so they can be installed anywhere. Nothing on the site said they existed, so the only way to find them was to already know. The page says what each skill carries and where each agent looks for a SKILL.md, because that is the part that differs: the file is the same in every case and only the directory changes. The Claude Code row is the one this repository can vouch for; the rest are stated from each tool's own documentation and the page says so, rather than implying every path was tried. It sits under Automate, next to the scripting pages, because the reader who wants it is the reader automating a session. Scripting, quality gates and the repository list link to it so it is reachable rather than orphaned. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This was referenced Sep 1, 2026
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
2 participants
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.
What
A new page, Agent skills, under Automate: what the three skills MeshBench maintains are for, and where each agent looks for a
SKILL.mdso they can be installed.Why
The skills live in
.claude/skills/in the application repository and are mirrored intoMeshBench/meshbench-scripting-skillsandMeshBench/meshbench-dev-skills. Both mirror repositories are already named in the repository list here, with no page saying what they contain or how to use them, so finding them required knowing about them first.The install section covers more than one agent deliberately. A skill is a directory holding a
SKILL.mdunder the Agent Skills open format; what differs between tools is only the directory, so the page is a table of directories rather than a set of separate procedures..agents/skills/is called out because Cursor, Gemini CLI, Codex and VS Code all read it, and a symbolic link into it installs once for several agents.Only the Claude Code row is verified, and the page says so in bold. Every other row is taken from that tool's own documentation, linked at the foot of the page, and was not exercised. Directories were read today from: Claude Code, VS Code, Cursor, Gemini CLI and Codex documentation.
Placement
Automate, immediately after the scripting cookbook. The reader who wants a skill is the reader automating a session, and prev/next then walks scripting, cookbook, skills, then the references.
Linked from
scripting.md(a new closing section),quality-gates.md(the interface rules are carried as a skill rather than as a check) andrepositories.md(the two mirror repository rows).Checks
python3 gen.pycompletes and reports 39 pages. The generated pages were current against the neighbouring checkout, so no sync script had to run. No em-dashes.