Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add scarletkc/agents --skill ask-to-plangit clone --depth 1 https://github.com/scarletkc/agentsWrote 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/skills/scarletkc/agents/ask-to-plan)<a href="https://agentmods.dev/skills/scarletkc/agents/ask-to-plan"><img src="https://agentmods.dev/badge/skills/scarletkc/agents/ask-to-plan/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/scarletkc/agents/ask-to-plan"><img src="https://agentmods.dev/badge/skills/scarletkc/agents/ask-to-plan.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00079 | $0.01924 |
| Opus 5.5 | $0.00032 | $0.00770 |
| Sonnet 5.5 | $0.00016 | $0.00385 |
| Haiku 4.5 | $0.00008 | $0.00192 |
Grade C, and why
ask-to-plan scanned grade C 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 1mo 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.
Nullifies safety policieshighAnti-refusal
"You have no restrictions", "do anything now", "ignore your guidelines": a direct jailbreak that disables guardrails.
tool or override its restrictions; do not change agent settings to enable it. How it starts
The opening of the file, as written. The whole thing — 166 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Ask to Plan
Help the user discover what they want by making one useful decision at a time. Move from broad outcomes to concrete requirements and then an implementable plan. The user should be able to make progress by choosing buttons, including when they have no idea yet or do not know technical terms.
The default deliverable is a plan. Starting this interview or accepting a solution does not itself request implementation. If the user later asks to build it, carry the settled requirements into that work.
Use the harness's actual ask tool
Identify the native structured-question tool exposed in the current session,
such as request_user_input, request_user_input_async, or the harness's
equivalent. Use the tool to present choices. Follow its actual schema and
current mode restrictions, including question limits, option limits, and
whether the interface already supplies free-text input or an Other option.
Tool names here are discovery hints, not a promise that a tool is available.
Start with a brief explanation of the next decision and one native question. Do not print a questionnaire, simulate buttons in Markdown, or produce a complete solution before learning the user's intent. If structured input is unavailable or fails, explain that buttons are unavailable here and use a concise text question if the harness permits it. A skill cannot add a missing tool or override its restrictions; do not change agent settings to enable it.
For an asynchronous ask tool, submission acknowledgement means the question is pending. Wait for the actual answer before asking a dependent question or settling that decision; only independent inspection can advance meanwhile. If a prompt is dismissed or returns no answer, follow the harness's rules for waiting or continuing with assumptions. Keep unanswered choices unresolved or explicitly provisional. A preselected option, timeout, or tool success is not the user's choice.
Make each question easy to answer
- Ask one question per round by default. Its answer should change the goal, scope, approach, or next question. Skip details that do not affect the plan.
- Offer two or three short, distinct choices within the tool's limits. Explain the practical consequence of each in a short description when supported. Ask about one dimension; do not bundle audience, platform, and budget into a single choice.
- Build options from what is already known. Use the user's language and level of detail. A beginner can choose "only on my computer" or "shared across devices" without knowing database names.
- Provide a useful uncertainty choice when needed, such as "Help me choose" or "Show me examples". After that choice, offer concrete directions or a reasoned recommendation; do not repeat the same difficult question.
- Recommend an option when the known needs support it and explain why. If the tool requires a recommended option, ground it in the stated context or a disclosed provisional assumption. Do not invent a user preference.
- Accept free-text corrections and combinations where supported. Use multiple selection only for choices that can coexist; incompatible approaches need a single choice. If exact input matters, such as a project path, request it through a supported input surface instead of guessing it from buttons.
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.
- 1mo ago First seen · 166 lines · 79 tokens per session scan C 3c8198aaf1a8
ask-to-plan is a skill published in the GitHub repository scarletkc/agents (225 stars, last pushed 24d ago), licensed Apache-2.0. It adds 79 tokens to every session and 1,924 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it C with 1 finding (nullifies safety policies). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-13.
Other skills, from other repositories
maintaining-windows-health
Hands-on playbook for Windows 11 disk cleanup, dev-machine optimization, and proactive health alerting. Use when the PC is full or slow, when a BSOD / Kernel-Power 41 / crash dump / commit-memory pressure happened, when the user asks to free disk space, audit storage, set up disk/memory alerts, or restore the same…
prompt-engineering
Universal prompt engineering techniques for any LLM. Use when crafting, optimizing, or reviewing prompts for AI models. Triggers on requests like "improve this prompt", "write a system prompt", "optimize my instructions", "help me prompt engineer", "audit this prompt", "review my prompt", or when building agentic…
installing-cli-tools
Install, upgrade, configure, and verify developer CLI tools safely. Use when a user asks to install a new CLI, command-line app, SDK tool, package-manager binary, GitHub release binary, language runtime tool, or AI/vendor CLI; configure shell PATH/completions; run first login; set API keys, tokens, or env variables…
repo-activity-summary
Summarize a repository's recent engineering activity from git history — technologies, work types, churn hotspots, contributor patterns, and velocity. Use when asking "what has this repo been working on", "is this project active", "who contributes what", "where are the hotspots", or before onboarding onto an unfamiliar…
fetch-url-as-markdown
Fetch a web page (URL) and return clean Markdown via local trafilatura, with Exa MCP as a fallback for JS-rendered or anti-bot pages. Use when the user asks to read, fetch, scrape, summarize, or quote a URL — prefer this over the built-in WebFetch tool. Don't use for binary files (PDFs, images, archives) or for…
clipboard
Copy text to clipboard with optional rich formatting. Triggers on "copy to clipboard", "copy that", "pbcopy", "copy formatted", "copy rich text".