viticci/shortcuts-playground-plugin is a Claude Code and Codex plugin that turns natural-language descriptions into signed Apple Shortcuts files for macOS and iOS. It helps users build, validate, and remix shortcuts by generating their XML and signing it with Apple’s shortcuts command-line tool. The catalogue entries are the plugin’s own agents, commands, skill, and hook.
Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add viticci/shortcuts-playground-plugin/plugin install claudeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/viticci/shortcuts-playground-plugin/build)<a href="https://agentmods.dev/commands/viticci/shortcuts-playground-plugin/build"><img src="https://agentmods.dev/badge/commands/viticci/shortcuts-playground-plugin/build.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00044 | $0.00837 |
| Opus 5 | $0.00022 | $0.00418 |
| Sonnet 5 | $0.00009 | $0.00167 |
| Haiku 4.5 | $0.00004 | $0.00084 |
Grade B, and why
build scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- **Never grep `~/Documents` broadly, `~/Agent`, `~/.claude/skills`, `~/.claude/plugins`, `~/.claude/projects`, `~/Library`, `/Applications`, or `/System`.** Even when an agent escalates on "I don't know this schema." Th How it starts
The opening of the file, as written. The whole thing — 33 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Delegate to the shortcut-builder agent (via the Agent tool with subagent_type: shortcut-builder) to design, build, validate, sign, and archive a macOS/iOS Shortcut that satisfies the following brief:
$ARGUMENTS
The agent owns the full workflow end-to-end. Your job as the orchestrator is minimal:
- Invoke the agent via the
Agenttool withsubagent_type: shortcut-builder, passing the brief as the agent's prompt. - Wait for the agent's result. The agent is responsible for: resolving the output directory, reading skill files, picking an icon, drafting, Craig-Loop validating, archiving, signing, verifying the signed file exists, and reporting absolute paths.
- Reject incomplete reports. If the agent returns only "validation passed", a draft XML path, or anything without a verified signed
.shortcutpath plus archive XML path, treat that as an incomplete build and re-delegate with: "Pipeline first: skip optional polish, run sign-shortcut on the current valid draft, verify the signed file exists, and report the signed and archive paths." - Relay the agent's report to the user verbatim. Don't re-run any steps the agent already completed.
When the agent escalates
The agent will escalate to you (the orchestrator) if it cannot build — typically because of a documentation gap on an allowlisted action, a validator error it can't fix in 5 iterations, or a brief it can't parse. When that happens:
- Do not do the agent's research for it. Do not grep
~/Documents,~/Agent,~/.claude,~/Library,/Applications,/System, or any other user or system path looking for examples or schemas. Those directories are off limits for this command. - The only paths you may read from are:
- The plugin installation directory itself (
${CLAUDE_PLUGIN_ROOT}— skill docs, bundled data, golden shortcuts, validator source). If the answer is here, read it and re-delegate to the agent with the answer as additional context. - The file the agent just wrote under the
OUTPUT_DIRit printed in step 0, if you need to inspect what it produced before responding.
- The plugin installation directory itself (
- Relay the escalation to the user if the plugin directory doesn't have the answer. Present the agent's escalation reason and any options it offered, and wait for user input. Do NOT paper over the escalation by improvising.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 33 lines · 44 tokens per session scan B 2a1eb11a2552
build is a command published in the GitHub repository viticci/shortcuts-playground-plugin (1,085 stars, last pushed 2mo ago), licensed MIT. It adds 44 tokens to every session and 837 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
lint-fix
Run all formatters and linters with auto-fix.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.