Brief Me

简体中文

Turn AI's wall of analysis into a consultant-style visual briefing you can explore, respond to, and decide from.

Don't dump the analysis. Brief me.

AI can generate more background, analysis, options, and risk assessment than a decision-maker can reasonably absorb. The person who has to make the call is still left rebuilding the storyline, finding the important relationships, and comparing the trade-offs across dozens of screens of text.

Brief Me turns that material into a consultant-style, interactive visual HTML briefing. It leads with what matters, makes the situation, data, and consequences visible, then lets the reader compare scenarios, change conditions, and express a view—supporting a higher-quality judgment.

Read the true origin story in Chinese.

From a wall of text to a real briefing

Long, complex analysis from AI, research, or planning
                         ↓ consultant synthesis
               A conclusion-first storyline
                         ↓ visual understanding
          Relationships, data, trade-offs, and impact
                         ↓ decision interaction
          Compare, change conditions, and respond
                         ↓
                  A current decision

Three capabilities

Consultant synthesis

Identify who must decide what, then select, rank, and organize the material like a well-prepared advisor. Put the conclusion, conflict, or trade-off with the greatest decision value first, and keep supporting detail available on demand.

Visual understanding

Choose comparison, relationship maps, data charts, causal chains, timelines, roadmaps, or scenario views from the actual cognitive task. Visual work also includes hierarchy, layout, and reading rhythm: its job is to reveal relationships that were hard to see in prose.

Decision interaction

Let the reader change the comparison, inspect support, adjust a condition, see the affected reasoning, and respond in their own words. When a change requires new model reasoning, the briefing captures the feedback and hands it back to the conversation.

What it creates

• A self-contained HTML briefing that opens directly and is easy to review or share.
• A consultant-style storyline designed for this decision rather than a table of contents for the source material.
• At least one visual treatment that materially lowers the cost of understanding.
• At least one meaningful interaction that gives the decision-maker more power than passive reading.
• A copyable summary of the current judgment and feedback for continuing the conversation.

One-line install

npx skills add jinhanbuilds/brief-me --skill brief-me -y

This package follows the Agent Skills specification. The command above uses the cross-host skills CLI; add --agent <host> when you need to select a supported host explicitly. For a manual install, copy skills/brief-me/ into a directory your host scans, such as ~/.agents/skills/brief-me/ where the shared convention is supported.

The core SKILL.md and its relative references are host-neutral. skills/brief-me/agents/openai.yaml and .codex-plugin/plugin.json are optional OpenAI/ChatGPT/Codex metadata; compatible non-OpenAI hosts may ignore them. Full delivery still requires a host that can create local HTML artifacts and open them for review.

Reopen the Agent host after installation, then invoke:

$brief-me Turn the long analysis above into a conclusion-first, interactive visual briefing that helps me make the decision.

Common requests

$brief-me Brief me on this: show me the whole situation, then compare the three directions.

$brief-me Turn this research into a decision briefing with data visuals and scenario switching.

$brief-me Present the analysis like a consultant and let me change conditions or add my own view.

The Skill may also be used when the user explicitly asks for a briefing such as “给我做个汇报” and both complex analysis and a real judgment task already exist.

Good inputs

Brief Me is strongest when the analysis already contains some combination of:

• a question or recommendation;
• background, research, data, options, or a working judgment;
• meaningful differences, relationships, trade-offs, or consequences;
• a real person who must decide, approve, change, or defer something.

If the material cannot yet support any judgment, Brief Me identifies what is missing instead of hiding the gap behind a polished interface.

Boundaries

• General summaries, pure status dashboards, and archives with no judgment task are outside its scope.
• Each briefing is designed from the current problem rather than forced into a fixed dashboard, matrix, or deck outline.
• It does not fabricate scores, probabilities, data, or dynamic calculations.
• It does not write a displayed choice back to a source system or treat feedback as authorization.
• It preserves source, date, unit, and uncertainty when they matter to the decision.

Generated HTML may contain sensitive source material. Review it before sharing or publishing.

Package contents

• skills/brief-me/ — the complete runtime Skill and design references.
• docs/product-story.zh-CN.md — the anonymized true origin and product promise.
• .codex-plugin/plugin.json — optional OpenAI/ChatGPT/Codex Plugin metadata.
• scripts/ and tests/ — package validation, maintainer-only local installation sync, and reproducible release building.
• skills/brief-me/evals/ — result-level behavior cases that inspect generated briefing artifacts.

Status and validation

Current release candidate: v0.1.0.

npm run validate
npm test
npm run build

Contributors with skill-up installed can also run npm run eval:validate and npm run eval:run. These cases inspect generated HTML rather than treating package structure as proof of product behavior.

Package validation establishes that the release is complete and installable. Result-level evaluation then asks the concrete product question: did the synthesis reduce cognitive load, did the visuals expose the important relationships, and did the interaction help the decision-maker form and express a judgment?

License

MIT © 2026 Jinhan