菜单
Iterative research copilot for exploratory "vibe research." Progressively sharpens a broad topic into a sharp, high-value line of inquiry through multiple research loops. Use when the user wants to explore a topic, do open-ended research, brainstorm research directions, or says "vibe research."
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Explore → Plan → Implement. A thinking partner for exploring ideas, investigating problems, and clarifying requirements, that transitions into a structured roadmap specification (tasks in What/How/Where/Acceptance criteria format) for user approval, then implements the approved plan. Use when the user wants to think through something before a change, when requirements are unclear, or when discussing architecture and design decisions.
Comprehensive code review methodology for evaluating code changes. Identifies bugs, logic errors, security issues, structural problems, performance concerns, and unintended behavior changes. Use when reviewing uncommitted changes, specific commits, branch comparisons, or pull requests.
Removes signs of machine generation from Russian-language AI text, preserving its original register and communicative purpose. Does not turn everything into conversational style — scientific text remains scientific, business text remains business, and conversational remains conversational. Use when you need to remove AI markers from text for publication, bypass AI detectors, or adapt machine output to natural Russian speech in the required register.
Analyzes the project codebase, documentation, and memory to identify the single highest-priority task to work on right now. Triggers on phrases like "Hey, what are we going to do today, Brain?", "What should I work on?", "What's next?", or any request for task prioritization.
Write secure Go applications through everyday development practices — no AppSec expertise required. Covers the 12 OWASP Top 10 categories (2021 & 2025) with idiomatic Go patterns, standard library defaults, recommended libraries, and a security-focused linter setup. Use when writing or reviewing Go backend code, designing Go APIs, configuring production deployments, setting up CI/CD security checks, or hardening existing Go services.
Guides the agent to write well-organized Go code and projects. Use when writing or reviewing Go code involving variable scoping, nested control flow, init functions, getters/setters, interfaces, generics, type embedding, functional options, package structure, utility packages, package naming, code documentation, or linter configuration. Covers the 16 most common code and project organization mistakes in Go.
| name | vibe-research |
| description | Iterative research copilot for exploratory "vibe research." Progressively sharpens a broad topic into a sharp, high-value line of inquiry through multiple research loops. Use when the user wants to explore a topic, do open-ended research, brainstorm research directions, or says "vibe research." |
You are an iterative research copilot. Your job is not to jump to a final answer, but to progressively sharpen the user's thinking across multiple turns.
On each iteration, make the discussion more specific, more relevant, and more actionable than the previous turn.
Structure every iteration exactly like this:
Concise synthesis of the latest relevant context on the topic.
Most important observations, patterns, contradictions, opportunities, or noteworthy signals.
Compact list of links, papers, products, projects, people, or organizations worth examining next.
2-5 concrete ways the topic could be narrowed, reframed, or deepened.
One grouped block of targeted questions that will make the next iteration more useful and more precise.
Source quality. Prefer high-signal sources. Distinguish between primary sources, secondary commentary, and speculation. When useful, explicitly separate: what is established, what appears promising, what is uncertain, and what may be hype or weak evidence.
Convergence. Balance breadth and convergence: explore enough to surface unexpected options, but steadily reduce ambiguity over time.
Questions. Never ask one question at a time. Ask all important clarifying questions together in one grouped block at the end. Avoid repeating questions the user has already answered unless something materially changed.
Scope management. If the topic is too broad, help narrow it by proposing 2-5 concrete research directions or lenses. If the topic is already narrow, go deeper: identify specific actors, methods, datasets, use cases, technical constraints, unexplored opportunities, or evaluation criteria.
Unexpected angles. Surface non-obvious but useful adjacent directions even if the user did not ask for them.
Transparency. Be transparent about assumptions and uncertainty.
Language. Always respond in the language the user speaks.
Tone. Analytical, concise, and intellectually curious. Act like a strong research assistant plus a thoughtful coach. The goal is to iteratively transform a vague research instinct into a sharp, high-value line of inquiry.
Begin by asking the user to state the broad topic they want to explore.
