meta — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited meta (Plugin) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
A curated set of Claude Code plugins that add structured workflows to your AI sessions — from Linear planning and QA, to billable hour tracking, to prose rewriting. Each plugin bundles several related skills you install once and use across every project.
Requires Claude Code v1.0.33 or later.
Some plugins reach outside Claude Code itself. Check what you need before installing:
| Plugin | External dependencies |
|---|---|
delivery | Linear MCP (all skills); Playwright MCP (/qa visual verification); OPENAI_API_KEY env var (/plan-design image generation) |
meta | None — runs entirely inside Claude Code |
specialist | None — connects to your project's database at runtime |
track | jq, git (both must be on your PATH) |
writing | None |
If you already know which plugin you want:
/plugin install delivery@mvdmakesthings
/plugin install writing@mvdmakesthings
/plugin install track@mvdmakesthings
/plugin install meta@mvdmakesthings
/plugin install specialist@mvdmakesthingsAdd the marketplace first, then explore from the Discover tab:
/plugin marketplace add mvdmakesthings/skillsThen open the plugin browser:
/pluginNavigate to the Discover tab and install what you need.
| Plugin | What it does |
|---|---|
delivery | Linear-driven planning → QA flow: PRDs, issue slicing, plan grilling, design mockups, test planning, QA execution, and post-incident reports |
writing | Humanize AI-sounding prose; storytelling coach for pitches, talks, and memos |
track | Billable hours timer backed by a git-versioned plaintext ledger |
meta | Improve your own skills: capture session friction, then evolve SKILL.md files with your approval |
specialist | Domain-expert personas: a Postgres-first DBA for audits, schema design, index/bloat cleanup, slow-query diagnosis, and live incident investigation |
delivery/plugin install delivery@mvdmakesthings
A full planning-to-QA pipeline built around Linear. Skills chain together — /to-prd → /to-issues → /plan-design + /plan-qa → /qa — but each one works independently.
| Command | What it does | Requires |
|---|---|---|
/to-prd | Synthesizes the current conversation into a PRD and publishes it to Linear | Linear MCP |
/to-issues | Slices a plan, spec, or PRD into independently-grabbable Linear issues — each a thin vertical slice with its own acceptance criteria | Linear MCP |
/grill-with-docs | Challenges a plan against your CONTEXT.md and ADRs one question at a time, resolves terminology conflicts, and writes decisions back into those docs inline | CONTEXT.md / ADRs |
/plan-design <issue-id> | Reads a Linear issue, interviews you per UI surface, generates three layout mockups via OpenAI image generation, builds HTML/CSS for the one you pick, and attaches it to the ticket | Linear MCP, OPENAI_API_KEY |
/plan-qa <issue-id> | Drafts a layer-aware test plan from a ticket's acceptance criteria and attaches it as <issue-id>-test-plan.md for /qa to execute | Linear MCP |
/qa [issue-id] | Maps acceptance criteria to coverage, runs every test layer, executes the attached test plan, and screenshots the UI against design attachments | Linear MCP, Playwright MCP |
/pir | Guides a post-incident retrospective: surfaces what broke, why, and what changes prevent recurrence — produces a structured PIR document | None |
writing/plugin install writing@mvdmakesthings
| Command | What it does | Requires |
|---|---|---|
/writing:human [prompt] | Rewrites prose so it reads like a person wrote it — strips the vocabulary and sentence patterns that flag text as AI-generated | None |
storyteller-guidance | A storytelling coach for pitches, talks, and memos. Auto-triggers on pitch / talk / memo phrasing. Draws from a 54-card tactic deck | None |
track/plugin install track@mvdmakesthings
| Command | What it does |
|---|---|
/track:start <client> | Start a billable timer for a named client |
/track:stop [note] | Stop the active timer with an optional session note |
/track:pause | Pause the active timer |
/track:resume | Resume a paused timer |
/track:status | Show the active timer and today's totals per client |
/track:report | Print an invoice-ready rollup with hours and dollar totals |
Sessions are recorded to a git-versioned JSONL ledger under ~/.time-tracker/. Requires jq and git on your PATH.
meta/plugin install meta@mvdmakesthings
Two skills that form a continuous improvement loop for any skill in this marketplace:
Any project Skills repo
─────────── ───────────
Run a skill
↓
/skill-reflect ──── logs ───→ ~/.claude/skill-sessions/
↓
/skill-improve ──→ SKILL.md updated
↓
HITL approval gate| Command | Where to run | What it does |
|---|---|---|
/skill-reflect <skill> | Any project, immediately after a skill session | Mines the conversation for friction — improvised steps, user corrections, missed edge cases — and writes a structured log to ~/.claude/skill-sessions/<plugin>/sessions/<skill-name>/ |
/skill-improve <skill> | Skills repo checkout only | Reads accumulated session logs, proposes numbered SKILL.md changes with evidence, waits for your approval on each, then applies accepted changes, bumps the patch version, and writes a CHANGELOG.md entry |
Reflect after sessions anywhere → logs accumulate in ~/.claude/skill-sessions/ → improve from this repo when you're ready.
specialist/plugin install specialist@mvdmakesthings
Domain-expert personas that drop into any project and discover its conventions at runtime. Currently ships one expert: a Postgres-first DBA.
| Command | What it does | Requires |
|---|---|---|
/dba | Runs a playbook chosen from: full health audit, schema design + implementation with tests, index/bloat cleanup, slow-query diagnosis and fixes, or live incident investigation | Postgres access |
The DBA skill auto-triggers on database-shaped prompts. It discovers your project's migration tool, test runner, and deploy conventions before acting, and defers rule-level Postgres detail to the supabase-postgres-best-practices skill when it's installed alongside this plugin.
.claude-plugin/
└── marketplace.json ← Marketplace catalog (lists all plugins)
plugins/
├── delivery/
│ ├── .claude-plugin/
│ │ └── plugin.json ← Plugin manifest
│ └── skills/
│ ├── to-prd/SKILL.md
│ ├── to-issues/SKILL.md
│ ├── grill-with-docs/SKILL.md
│ ├── plan-design/SKILL.md
│ ├── plan-qa/SKILL.md
│ ├── qa/SKILL.md
│ └── pir/SKILL.md
├── writing/
│ └── skills/
│ ├── human-voice-writer/SKILL.md
│ └── storyteller-guidance/SKILL.md
├── track/
│ ├── bin/track.sh ← Bash dispatcher (all ledger logic lives here)
│ └── skills/
│ └── track/SKILL.md
├── meta/
│ └── skills/
│ ├── skill-reflect/SKILL.md
│ └── skill-improve/SKILL.md
└── specialist/
└── skills/
└── dba/SKILL.md
_dev/ ← Dev-only: tests, evals, scratch docs. Not shipped on install.Test a plugin locally without publishing:
# Add your local clone as a marketplace source
/plugin marketplace add ./path/to/skills
# Install the plugin you're working on
/plugin install delivery@mvdmakesthings
# Validate the full marketplace
claude plugin validate .Create a SKILL.md under the relevant plugin — no marketplace change needed:
plugins/
└── delivery/
└── skills/
└── your-skill/
└── SKILL.mdPut dev-only fixtures (tests, eval data) under _dev/<plugin>/ so they don't ship on install.
mkdir -p plugins/your-plugin/.claude-plugin// plugins/your-plugin/.claude-plugin/plugin.json
{
"name": "your-plugin",
"description": "What your plugin does",
"version": "1.0.0"
}plugins/your-plugin/skills/<skill-name>/SKILL.md..claude-plugin/marketplace.json:{
"plugins": [
{
"name": "your-plugin",
"source": "./plugins/your-plugin",
"description": "What your plugin does"
}
]
}claude plugin validate .See the Claude Code plugin docs for details on skill manifests, agents, hooks, and other component types.
MIT
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.