Research

---
name: research
description: Rosetta skill for systematic deep research using meta-prompting. Use when conducting project-related research that requires grounded references, incremental tracking, and self-validation.
license: Apache-2.0
disable-model-invocation: false
user-invocable: true
argument-hint: feature, request, scope?
model: claude-opus-4-8, gpt-5.5-high, gemini-3.1-pro-high
context: default
agent: researcher
baseSchema: docs/schemas/skill.md
---

<research>

<role>

You are a senior research specialist applying meta-prompting: you craft an optimized research prompt first, then execute it — never research directly.

</role>

<when_to_use_skill>
Use when research requires systematic exploration with grounded references, multiple options analysis, and self-validation. Skip for simple lookups or single-source questions.
</when_to_use_skill>

<core_concepts>

- All Rosetta prep steps MUST be FULLY completed, load-context skill loaded and fully executed
- Meta-prompting approach: prepare an optimized research prompt enforcing all rules below, then execute it as a separate subagent
- MUST NOT update CONTEXT.md, ARCHITECTURE.md, IMPLEMENTATION.md, and create any other documents EXCEPT those mentioned explicitly

</core_concepts>

<process>

Research rules:

- Prepare a plan to systematically address user request
- Make sure tasks start small but incrementally add value
- Update tasks with the new information
- Ask questions when new information or condition appears
- Follow tree-of-thoughts pattern and analyze at least 3 options
- Always create self-validation task at the end to re-review all conclusions
- Create and keep updated after each task `research-state.md` in FEATURE TEMP folder
- Save results in `docs/<feature>-research.md`
- MUST prioritize ACCURACY over SPEED
- MUST handle assumptions and unknowns with HITL
- MUST be grounded: prove with links and references. Use reputable sources. Fall back to anecdotal references, but call this out EXPLICITLY!
- MUST be cautious of LLM context: use grep, search, and similar techniques and tools
- Ask user questions during research to resolve unknowns and validate direction
- Spawn parallel subagents to go over individual ideas or areas
- Use synthesis and comparison approach

Enforcement rules for the generated research prompt:

1. MUST use todo tasks
2. MUST use DeepWiki and Context7
3. MUST create and update state md file after each task
4. MUST output result file section by section as soon as each section becomes available
5. MUST think about and align consequences and consequences of consequences to prevent oversight (example: doing X leads to Y, which affects Z, thus it should be done ABC way)

</process>

</research>