jons-ai-toolkit
A Claude Code plugin marketplace bundling skills, subagents, and slash commands across six themed plugins.
Install
Add the marketplace, then install whichever plugin(s) you want:
/plugin marketplace add jon-the-dev/jons-ai-toolkit
/plugin install everything@jons-ai-toolkit # full kit
/plugin install dev-agents@jons-ai-toolkit # or pick themed plugins
For local development against a clone:
/plugin marketplace add /path/to/jons-ai-toolkit
Plugins
react-core-workflow
Skills for working in the React core monorepo.
extract-errors — run yarn extract-errors and report missing error codesfeature-flags — manage React feature flag files, gating tests with @gate pragmasflags — inspect feature flag state across release channels (source, www, stable, experimental)flow — run Flow type-checking against a specific renderer (dom-node, dom-browser, native, fabric)fix — run yarn prettier and yarn linc to clean up before committingtest — run React tests across release channels with channel/variant flagsverify — run the full pre-commit verification chain (prettier, linc, flow, tests)
gh-issue-workflow
GitHub issue triage and execution loop.
Commands:
/show-backlog — fetch all open issues and render a P0–P7 prioritized backlog table/next-issue — recommend the single next issue to work on, applying security > bug > feature > perf priority/implement — take a single issue end-to-end: branch, scan, delegate to dev subagents, test, PR, close/mini-sprint — pick 3–4 issues from the backlog and run them in parallel via subagents
Skills:
gh-cli — comprehensive gh CLI reference (repos, issues, PRs, Actions, releases, codespaces)
dev-agents
General-purpose development subagents.
backend-developer — building application backends (FastAPI, FastMCP, Python)frontend-developer — building UIs and React/Vue/Angular components, optimizing perf and a11ycloud-architect — designing AWS/Azure/GCP infra, Terraform IaC, cost optimization, multi-regionsecurity-auditor — comprehensive security audits, fans out OWASP, auth, secrets, deps, API, IaC, crypto subagentstest-runner — running tests across Python and JS/TS frameworks
pkm-toolkit
Personal knowledge management.
obsidian — read/write/search an Obsidian vault via CLI (notes, daily notes, tags, tasks, frontmatter)task-management — lightweight TASKS.md-based task trackingmemory-management — two-tier memory system (CLAUDE.md + memory/ directory) for workplace context and shorthandfind-skills — help users discover and install skills from the open ecosystem
content-toolkit
Content and operations.
content-creation — draft marketing content (blog posts, social, email, landing pages, press releases, case studies)data-visualization — choose and build effective charts in Python (matplotlib, seaborn, plotly)realtime-alerts-skill — design Slack and Microsoft Teams operational alert messages with structured interview
everything
A bundle that installs every skill, agent, and command in this repo. Use this for the full kit; install individual themed plugins for narrower scope.
Repo layout
.claude-plugin/marketplace.json # marketplace manifest
plugins/
<plugin-name>/
.claude-plugin/plugin.json # per-plugin manifest
skills/ -> ../../../skills/... (relative symlinks)
agents/ -> ../../../agents/...
commands/ -> ../../../commands/...
skills/ # canonical source of truth
agents/
commands/
Top-level skills/, agents/, and commands/ are the source of truth. Each plugin curates a subset via relative symlinks, so a single skill (e.g. gh-cli) can appear in multiple plugins without duplication. Symlinks are preserved by Claude Code's plugin cache per the path traversal limitations docs.
Versioning
Plugins are unversioned for now — installs resolve against this repo's git HEAD. Versioned releases will land once the API surface stabilizes.
jon-the-dev
