toolkit-update
Skill2 starsChanged 4 months agoMITView on GitHub
Compare, plan, or apply Portable Web Toolkit or toolkit-skill updates. Use when asked to inspect freshness, reconcile a local export with GitHub, inspect installed skill versions, or review an update. Normal toolkit commands perform a guarded, throttled freshness preflight; this skill handles manual review and reconciliation.
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- Web_Toolkit/AGENTS.mdAGENTS.md
- brand-doctorSkill
- discovery-doctorSkill
- image-pipelineSkill
- instagram-cloneSkill
- pagespeed-diagnosticsSkill
- portable-web-toolkit-routerSkill
- portable-web-toolkitSkill
- site-maintenanceSkill
- site-onboardingSkill
- site-readinessSkill
- site-starterSkill
- vectorize-pipelineSkill
- wcag-auditorSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
Read VERSION, package versions, skill-pack.json, Git status, current commit and remote, when available.
- Distinctive: line 27
Before reporting a Git remote, remove URL userinfo, credentials, query strings, and fragments.
- Distinctive: line 31
Use the optional status helper only as an aid:
- Distinctive: line 41
Prefer an upstream-backed Git worktree.
- Distinctive: line 43
create a recoverable backup
- Very specific: line 50
Never claim that git pull updated an export without .git.
- Distinctive: line 61
run an unaudited remote installer
- Distinctive: line 64
delete a local-only module merely because upstream lacks it
- Very specific: line 75
After a toolkit git pull, refresh the accessibility engine without bumping toolkit VERSION:
- Distinctive: line 88
When upgrading an existing install to v0.3.2+:
- Very specific: line 91
Run node scripts/manage-project-skills.mjs link --project .
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Toolkit Update
Establish current state before choosing an update.
Preflight
Resolve the active repository and applicable instructions, preserve unrelated work, then locate the actual toolkit source and selected client skill roots.
Inspect
- Resolve the toolkit root and determine whether it is a real Git worktree, an export, a cache, or a linked client copy.
- Read
VERSION, package versions,skill-pack.json, Git status, current commit and remote, when available. - Inspect the selected upstream release, tag, or immutable commit.
- Compare skill directories, versions, hashes, toolkit modules, templates, instructions, and release notes.
- Classify local-only, upstream-only, modified, generated, and ignored content.
- Report the status. The automatic preflight may apply only the guarded fast-forward and WCAGate refresh described below; manual reconciliation still stops unless explicitly authorized.
Before reporting a Git remote, remove URL userinfo, credentials, query strings, and fragments. Describe private-fork locations generically unless the exact locator is necessary and authorized.
Use the optional status helper only as an aid:
node ./scripts/check-agent-skills.mjs --agent <client> --scope <user-or-project>It must not be required for an agent-driven installation.
Plan
Prefer an upstream-backed Git worktree. For an exported local folder:
- create a recoverable backup
- clone the selected upstream source into staging
- port reviewed local-only capabilities without dependencies or caches
- resolve instruction and manifest changes
- validate in staging
- switch the active path only after review
Never claim that git pull updated an export without .git.
Apply
For manual source reconciliation, require exact update authorization. Use
git pull --ff-only only when the worktree and branch permit it. Otherwise use
a staging clone and explicit reconciliation. Normal command-time freshness uses
the same clean-main/fast-forward gate and never overwrites local changes.
Do not:
- run an unaudited remote installer
- overwrite unmanaged or locally modified skills
- reinstall every client as a side effect
- delete a local-only module merely because upstream lacks it
- install host software, push, tag, publish, or deploy without separate approval
WCAGate engine (independent of toolkit VERSION)
The toolkit automatically checks the unpinned public main head at most once
per 24 hours. It refreshes the ignored Web_Toolkit/node_modules install when
the remote head changes and writes the result to
Web_Toolkit/.runtime/toolkit-auto-update.json. Set
PORTABLE_TOOLKIT_AUTO_UPDATE=0 to disable the preflight.
After a toolkit git pull, refresh the accessibility engine without bumping toolkit VERSION:
cd Web_Toolkit
npm run wcagate:update
node ./wcag_auditor/bin/wcag-auditor.mjs core-path
# → …/Web_Toolkit/node_modules/@imyourboyroy/wcagateOr node ./Web_Toolkit/wcag_auditor/bin/wcag-auditor.mjs doctor --update-engine. That reinstalls the GitHub spec (npm update does not refetch unpinned git deps). npm 12+ needs --allow-git=all (shipped in Web_Toolkit/.npmrc). Do not clone AI/wcag-auditor.
Upgrading to Dynamic Symlinked Skill Architecture
When upgrading an existing install to v0.3.2+:
- Purge legacy heavy web skills (
instagram-clone,portable-web-toolkit,site-readiness,site-starter,toolkit-update,vectorize-pipeline) from global home skill directories (~/.gemini/config/skills/,~/.cursor/skills/,~/.claude/skills/). - Copy
skills/portable-web-toolkit-routerinto global home skill discovery directories. - Run
node scripts/manage-project-skills.mjs link --project . --skills portable-web-toolkit,site-readiness,site-starter,toolkit-updateto symlink selected skills into active project scope (.agents/skills/). Barelinkwithout--skillsonly installs the lightportable-web-toolkit-router.
Verify
Run:
- skill manifest validation
- structural validation for every skill
- positive and near-miss routing tests when available
- toolkit verification and privacy checks
- status comparison for selected installed clients
Report the previous and resulting versions, source commit, preserved local capabilities, files changed, installed skills left untouched, checks, rollback location, and unresolved publication work.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

