Research
Stop Claude from winging it — 7 workflow skills that route every task through two thinking modes (Philosopher = absorb / Engineer = ship). Works in Claude Code + Codex.
npx -y skills add Lemonaderrrrr/flow-kit --skill researchAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Turn a question or topic into an evidence-backed, cited research output (literature review, technical report, option comparison, project scoping). Default Engineer mode (define question -> multi-source search -> cross-verify -> structured output); Philosopher mode for truly mapping a field's threads and debates. Can call the deep-research skill for heavy fan-out. Saves to outputs/research/. Use when the user says "research X", "do a survey of", or invokes the skill.
SKILL.md
2.4 KB, as published. Nobody here has run it
Research Workflow
Turn a question / topic into an evidence-backed, cited, deliverable output.
Start
- Route-confirm (see
CONVENTIONS.md): one line stating this is the research workflow and the suggested mode; wait for the go-ahead. - Confirm mode: default ⚙️ Engineer; the user may switch to 🧠 Philosopher.
- Decide on heavy search: for broad questions needing multi-source fan-out + fact-checking, call the
deep-researchskill; for small ones, use plain web search/fetch.
⚙️ Engineer mode (default)
Goal: a report you can act on or decide from, time-boxed.
- Define the question — The specific question. For whom, what decision? Bound the scope.
- Multi-source search — Papers / authoritative sites; cite every claim. Go deep-research if broad.
- Distill + cross-verify — Cross-check key conclusions across ≥2 sources; flag what's uncertain; never pass guesses as fact.
- Structured output — Review / comparison table / report, conclusions first, with citations.
- Capture — Save to
outputs/research/research_<english-slug>.md(conclusion + evidence + citations + next steps).
🧠 Philosopher mode
For truly understanding a field (not just shipping a report): map the threads, schools, and core disagreements, then form your own judgment.
- Map the field — Main schools / methods / timeline; who disagrees with whom and why.
- Trace to first principles — Each school's core assumptions; when do they hold or fail?
- Form a judgment — Where you stand and why.
- Capture — Same location.
Wrap-up conventions
- Live-search principle (mandatory) — all material searched live; citations must be traceable; don't fabricate sources.
- Separate fact / inference / opinion — don't blur them.
- Promote: if it's directional/reusable, offer to save it to long-term memory; one-off outputs stay put.
- Filenames use English slugs.