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 Jewgah/claude-code-skills --skill choicesgit clone --depth 1 https://github.com/Jewgah/claude-code-skillsWrote 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/jewgah/claude-code-skills/choices)<a href="https://agentmods.dev/skills/jewgah/claude-code-skills/choices"><img src="https://agentmods.dev/badge/skills/jewgah/claude-code-skills/choices/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/jewgah/claude-code-skills/choices"><img src="https://agentmods.dev/badge/skills/jewgah/claude-code-skills/choices.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 | $0.00047 | $0.01315 |
| Opus 5.5 | $0.00019 | $0.00526 |
| Sonnet 5.5 | $0.00009 | $0.00263 |
| Haiku 4.5 | $0.00005 | $0.00131 |
Grade A, and why
choices scanned grade A with 0 findings 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 yesterday.
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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Choices - WhatsApp decision summary
Generate a concise, copy-paste WhatsApp message (English) that lists the open
decisions the team needs to make, based on the current conversation. This is the
sibling of /explain-dev (which announces finished work): /choices
surfaces pending decisions in English so the team can pick.
The audience is ALWAYS the client
Every /choices message is written to be forwarded to the CLIENT (the partner or
stakeholder who pays for the work), never to an internal dev channel. Write it as
if the client is reading every line, because they are. Two consequences that are
easy to get wrong:
- Address the client in the second person. If a named person owes an answer and that person is the recipient, write "you answered 1 to 4 in July but not this one", never "Alex answered 1 to 4". Third-personing the reader is an instant tell that the message was written for someone else.
- Money questions: ask what is theirs to decide, never what is mine. The line
is who the decision belongs to, not whether money is mentioned.
- Legitimate (the client genuinely decides): approving a quote for upcoming work, which period to bill something to, whether to authorise more spend, how to prioritise a fixed envelope. Put these in plainly, with a number when I have one.
- Never (mine alone): whether I gift, absorb, discount or write off work, my margin, my capacity, my internal hour estimates. "Should I gift you the overage?" is nonsense addressed to the person paying.
- When the client has said "build everything", a scope question is not a choice, it is a quote. Do not re-offer a phased menu they already declined. State what will be built and give the number, framed for approval.
The test for every line: would I be comfortable if the client screenshotted this line back to me? If not, it belongs in my reply to the user, not in the message.
Steps
- Scan the conversation for every UNRESOLVED choice a human must make: forks, trade-offs, assumptions awaiting confirmation, "we could do X or Y", and any decision you were about to ask (e.g. an AskUserQuestion you queued). Include the scope/"how much to build" decision when relevant.
- For each, distil to a plain-language either/or (or 2-3 short options). Strip the technical detail - frame each by user/business impact, not implementation.
- Merge duplicates; order by importance (the choice that most changes the build goes first).
- Add a one or two line plain-language context at the top so the message stands alone for someone who wasn't in the conversation.
- End by inviting their answers.
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.
- yesterday Changed 37a463ed3b71
- 1mo ago First seen · 106 lines · 47 tokens per session scan A 22ede67c6047
choices is a skill published in the GitHub repository Jewgah/claude-code-skills (2 stars, last pushed 2d ago), licensed MIT. It adds 47 tokens to every session and 1,315 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-14.
Other skills, from other repositories
fable-validate-loop
Use when the user asks to validate a GitHub issue with Fable 5.1 and then autonomously drive it to a reviewed PR in one shot — "fable-validate-loop", "fable validate and work on #N", "fully automate issue #N with fable". Runs fable-validate, auto-applies its update-issue edits when the verdict calls for it, has…
mx-flow
Full development workflow orchestrator: brainstorm → plan → worktree → convergent TDD/review loop → verify → PR. One human gate (spec approval); all other gates auto-proceed. Use when starting a feature or significant change from scratch. Usage: /mx-flow ; /mx-flow status [name]; /mx-flow finish .
exam-prep
An exam-review skill for making quick-reference study materials such as formula cards, key-point summaries, mistake lists, and answer templates.
review-plan
A study-plan maker for exam preparation. It turns an exam date or a requested number of remaining days into a structured review schedule.
github-issue-pr-attribs
Standardize GitHub ISSUE and PR metadata (Assignee, Reviewer, Labels, Project, Milestone, Development, Relationship) and ship a strong, well-structured code-review template. Includes ready-to-use issue forms, PR template, and a severity/rubric review model.
triage
Move issues and external PRs through triage roles — categorise, verify, grill if needed, write agent-ready briefs.