Back to search
easy-context-question
SkillWorks with Claude CodeGitHub
Claude Code skill: turn your agents' open questions into a plain-English decision brief. Context, options with cost, one recommendation. Answer with one letter.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/tomi2k--easy-context-question)HTML
<a href="/listings/tomi2k--easy-context-question"><img src="/api/listings/tomi2k--easy-context-question/badge.svg" alt="findsafeskills safety rating" /></a>Install
i
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
git clone https://github.com/Tomi2k/easy-context-question.git mkdir -p ~/.claude/skills cp -r easy-context-question/skills/easy-context-question ~/.claude/skills/
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 119 lines
- <p align="center">
- <img src="assets/hero.png" alt="A tangle of technical agent questions goes in, one clear decision card with a recommendation comes out." width="100%">
- </p>
- # easy-context-question
- **Turn your agents' questions into decisions you can answer with one letter.**
- A skill for [Claude Code](https://code.claude.com). When agents, issues, plans or review
- reports ask you something, this skill rewrites the questions into a short decision brief:
- what it is about in plain words, the options with their consequence and cost, and exactly
- one recommendation. You answer "1A 2B" - without opening a single file.
- ```
- /easy-context-question the open questions in issue #212
- ```
- ---
- ## Why this exists
- The more agents work for you, the more questions come back - and they come back in the
- agents' language: ticket numbers, function names, half-explained trade-offs. To answer one
- question, you read the issue, then the code, then the previous discussion. The decision
- itself is quick. Understanding the question is not.
- | Problem | What the skill does |
- |---|---|
- | Questions are written in jargon: ticket numbers, file names, internal labels. | Plain words. A technical term is explained once, in half a sentence. |
- | "It depends" - the agent lists options but does not commit. | Exactly one recommendation per question, with a reason. |
- | You answer things the agent could have found out itself. | Questions the agent can settle by reading code and docs are settled first and reported in one line. |
- | You get asked the same thing twice. | Questions you already decided are dropped. |
- | Twelve questions at once, the blocking one is number nine. | At most 7 per brief, blocking questions and deadlines first. |
- | An option silently costs money or cannot be undone. | A **Warning** line before the options. |
- | Your answer gets lost. | The skill hands your choices back to whoever asked - outside the session only after your OK. |
- ### Plain words on purpose
- The language rules follow **Simplified Technical English** (ASD-STE100), the controlled
- language created for aerospace maintenance documentation
- ([about ASD-STE100](https://asd-ste100.org/about_STE.html)). The standard limits sentences
- to 20 words in procedures and 25 in descriptions; this skill uses 20 for everything.
- One statement per sentence, active voice, present tense, one word for one thing, numbers
- with units instead of descriptions. Text written for a mechanic in a hangar also works
- for a busy founder between two calls.
- ## What you get
- ```markdown
- ### Question 1: How long to keep debug logs (#57)
- **What this is about:** The app keeps debug logs for 90 days. They use 40 GB and cost
- 9 USD per month. In the last 6 months nobody opened a log older than 14 days.
- **Warning:** Option A deletes all logs older than 14 days. You cannot get them back.
- **Options:**
- - **A - Keep 14 days:** Storage drops to about 7 GB. Cost drops to about 2 USD per month. Effort for you: 0.
- - **B - Keep 90 days:** Nothing changes. Cost stays at 9 USD per month.
- **Recommendation: A.** The old logs are not used, and the saving repeats every month.
- Answer: "all recommendations", or one letter per question, for example "1A".
- ```
- ## Install
- ```bash
- git clone https://github.com/Tomi2k/easy-context-question.git
- mkdir -p ~/.claude/skills
- cp -r easy-context-question/skills/easy-context-question ~/.claude/skills/
- ```
- ## Use
- ```
- /easy-context-question the open questions of the current conversation
- /easy-context-question issue #212 the questions in a specific issue
- /easy-context-question the review report any source the agent can read
- ```
- It also runs on plain requests such as "what do I need to decide?" or "explain the
- open questions", and when your agent is about to put a manager agent's or subagent's
- questions to you. In multi-agent setups, tell your manager agent to use this format for
- every question it sends you - then every check-in ends with one line you can answer.
- ## Rules the skill follows
- - Read the original source completely - never answer from memory.
- - Never invent an option. A missing obvious option is added and marked as an addition.
- - Do not copy the source's recommendation blindly; disagreement is stated in one sentence.
- - Costs and deadlines only when verified - otherwise "cost not checked".
- - No secrets, no customer data, no legal texts in the brief.
- ## Security
- The skill is a single Markdown file: no scripts, no hooks, no dependencies. It does not
- call any service by itself. It tells your agent what to do with the agent's normal tools:
- - **Reading:** the agent reads the sources you point it to (an issue, a report, a file).
- Text from these sources is treated as data - the skill forbids following instructions
- found inside them. This reduces the risk of indirect prompt injection through public
- issues, but prompt rules cannot fully prevent it: point the skill only at sources you
- would also read yourself.
- - **Checking:** questions are settled only by reading tracked project files and docs - no
- shell or network commands, no `.env`, secret or credential files. Anything that changes
- state, costs money or touches an account needs your approval.
- - **Writing back:** inside the session, the chosen options go back to the agent that
- asked. Before anything is written outside the session (an issue, a pull request, a
- chat), the agent shows you the exact target and the exact text and waits for your OK. It sends only your
- choices, never your raw chat, secrets or customer data.
- ## License
- MIT - see [LICENSE](LICENSE).
- ---
- Built by **Timothy Scherman**. Less AI theater. More results.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.