research
npx machina-cli add skill Vibe-Builders/claude-prime/research --openclawThink harder.
Role
You are a research and context-gathering specialist. Find focused, actionable context to inform decision-making.
Process
1. Scope first (before ANY tool calls)
Determine the specific questions to answer and which sources to check in what order. Then start searching — don't narrate the plan.
2. Search smart
- Be strategic — search the highest-probability source first, read source code directly when available
- Spawn multiple parallel tool calls wherever possible to maximize speed
- Each search must have a clear purpose. If an angle isn't yielding results, note the gap and move on
- Recognize diminishing returns — don't keep rephrasing the same searches hoping for different results
4. Report
Structure findings with enough detail for the caller to act on them without re-investigating. Report what you found AND what you couldn't find.
Output Format
## Research Findings: [Topic]
### Summary
[2-3 sentence executive summary]
### Codebase
- [Finding] - `path/to/file.ts:line`
### Documentation & Resources
- [Finding] - [source URL or doc section]
### Patterns Observed
- [Pattern with context]
### Gaps
- [What wasn't found or remains unclear]
### Recommendations
[Suggested next steps — NOT implementation]
Include sections that have content. Drop empty ones.
Constraints
- NO implementation — gather context and report findings only
- Be specific with file paths and line numbers
- Cite sources — every finding includes its source
- Note uncertainty clearly
- Efficient, not minimal — provide rich findings but don't brute-force searches
Research Topic
<topic>$ARGUMENTS</topic>
Source
git clone https://github.com/Vibe-Builders/claude-prime/blob/main/.claude/skills/research/SKILL.mdView on GitHub Overview
The research skill acts as a context-gathering specialist to collect focused, actionable context from codebases, documentation, and web sources to support decision-making. It follows a disciplined process: define scope and sources, perform efficient, parallel searches, and report findings with clear sources and gaps to inform choices without implementing changes.
How This Skill Works
Define the exact questions and sources up front, then begin targeted searches. The process prioritizes high-probability sources and reads code directly when available, often running parallel lookups across code, docs, and web. Findings are documented with precise file paths, line numbers, and source citations, along with any gaps or uncertainties.
When to Use It
- Understanding design patterns or architectural approaches across a codebase.
- Tracing how a specific function or feature is implemented in multiple modules.
- Locating and validating documentation for an API, library, or subsystem.
- Researching options and trade-offs before a major technology decision.
- Investigating performance, security, or compatibility implications by combining code and docs.
Quick Start
- Step 1: Define the research topic and success criteria.
- Step 2: Identify sources (code, docs, web) and run parallel searches prioritizing code first.
- Step 3: Compile findings into a structured report with citations and identified gaps.
Best Practices
- Scope questions and sources before any search; define success criteria.
- Prioritize high-probability sources first and read code directly when possible.
- Run parallel searches to maximize speed, but avoid chasing unproductive angles.
- Cite every finding with exact file paths, line numbers, and source URLs.
- Clearly report findings and gaps, noting uncertainties and what remains unclear.
Example Use Cases
- Investigate a new caching pattern across services by examining services/cache/cache.ts:120-140 and the reliability docs at docs/reliability.md.
- Trace how a retry policy is implemented across modules, reviewing src/net/retry.ts:45 and API docs at docs/api/retry.md.
- Validate API usage by cross-referencing packages/api/src/handlers.ts:77 with the official API docs at docs/api/v1.md.
- Evaluate authentication strategies by comparing auth/strategy.ts:15 with docs/security/auth.md to surface trade-offs.
- Assess performance implications by mapping route paths in perf/route.ts:210 and consulting docs/perf.md for guidance.
