Evidence-Based Research
Investigates open questions and returns sourced, verified findings instead of assumptions.
- Repository
- Yeachan-Heo/oh-my-claudecode
- Stars
- 40,512
- Licence
- MIT
- Files
- 1 file
What it does
When a task depends on something the agent does not yet know, this skill tells it to stop guessing and investigate first. The agent searches the project's own code and docs, consults official documentation for external libraries or APIs, and checks more than one angle so it does not miss something. It then reports findings with a source for each, separates what it verified from what it inferred, flags contradicting evidence, and says what remains uncertain. It is designed to answer questions without changing any code.
Great for
- Developers trying to understand an unfamiliar codebase before making changes
- Anyone who needs a sourced answer about how an SDK, framework, or API actually behaves
- Teams that want a clear list of open questions and a recommended next step before committing to a plan
Skip it if
- Tasks where you just want code written or changed, since the skill explicitly forbids implementing as a side effect of research
- Quick factual questions the agent can answer from general knowledge, where the research overhead adds little value
Try asking
- “Research how our authentication middleware is called across the codebase before I refactor it”
- “Look up the official docs and tell me whether this API version supports streaming responses”
- “Investigate why this test started failing after last month's changes and cite the commits involved”
- “Find every place we read the user settings file and report any contradictions in how they're parsed”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
research.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip research.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip research.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- State your question precisely and say when it counts as answered, so the agent knows when to stop.
- Ask for external research explicitly if you want it to check official documentation rather than relying on the repo alone.
- Review the cited sources yourself for anything high-stakes, since the agent's citations are only as reliable as the documents it reads.
What's inside
- SKILL.md
Redistributed unmodified from Yeachan-Heo/oh-my-claudecode under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.