research-planning
Skill1.5k starsChanged 4 days agoMITView on GitHub
Use Gemini (or similar) deep research to create or refresh external field guides under README/ before planning or implementation. Use when the user wants exhaustive web research on a technology, API, framework, platform, or domain with practical examples and failure modes. Do not use for repo-only architecture work, local bug reproduction, or direct implementation without external research.
We split this skill into the rules it gives an agent
9 rules. 0 are common, 5 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 10
Adapted from Gist: Use when gemma-macos-tuner needs external research before planning or implementation.
- Distinctive: line 26
Read references/index.md first.
- Distinctive: line 28
Prefer updating an existing doc over creating a near-duplicate.
- Distinctive: line 29
Run research using whichever tooling exists:
- Very specific: line 30
If the repo adds a helper script (compare gist: scripts/gemini-deep-research.js), use it and the paths it expects.
- Distinctive: line 33
Add cross-links from README.md or related docs when appropriate.
- Distinctive: line 38
Use the section order and constraints from the gist copy of this skill if you need a long-form template (audience: senior engineers.
- Very specific: line 42
research-debugging: when the work narrows to a concrete external failure hunt. - Very specific: line 43
create-plan: when the guide unblocks implementation scoping.
“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Research Planning
Purpose
Adapted from Gist. Use when gemma-macos-tuner needs external research before planning or implementation. The durable artifact is usually a markdown file under README/ (often README/guides/ for long guides), not a chat-only summary.
Use When
- The user wants a new or refreshed guide doc.
- The work depends on current vendor, SDK, API, or platform behavior.
- The guide must answer what the thing is, how it works in practice, what breaks, and how to choose a path.
Do Not Use When
- The task is concrete implementation with enough local context.
- The task is a purely local bug with a live repro.
- The output should be temporary scratch instead of durable research.
Procedure
- Read references/index.md first.
- Lock scope: topic, versions, target path under
README/, and what decision the guide unblocks. - Prefer updating an existing doc over creating a near-duplicate.
- Run research using whichever tooling exists:
- If the repo adds a helper script (compare gist:
scripts/gemini-deep-research.js), use it and the paths it expects. - Otherwise: structured web research + Context7 for library facts + synthesis into
README/guides/<path>/<slug>.mdor another agreed location.
- If the repo adds a helper script (compare gist:
- Review and edit the draft: strip filler, verify versions and links, separate vendor facts from repo-specific integration notes.
- Add cross-links from
README.mdor related docs when appropriate. - Hand off to
create-planwhen the guide is clean enough to scope work.
Prompt contract
Use the section order and constraints from the gist copy of this skill if you need a long-form template (audience: senior engineers; exhaustive; code examples; Works Cited). Replace “Gist repo” with gemma-tuner-multimodal and target paths with README/....
Handoff Rules
research-debuggingwhen the work narrows to a concrete external failure hunt.create-planwhen the guide unblocks implementation scoping.
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- lingzhi227/agent-research-skills★ 350same name
- JunMA98/Computer-science-claude-skills★ 3same name
- dongzhigang13305312738-art/paper-skills★ 1same name
- sivaguru-2113/Agent-persona★ 1same name
- willlee1995/skills★ 1same name
- stritar/skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository20 other files it gives its agents.
- audit-fix-hardcoreSkill
- audit-fix-loopSkill
- auditSkill
- build-vs-buySkill
- create-planSkill
- create-skillSkill
- debugSkill
- deploySkill
- documentationSkill
- elon-muskSkill
- execute-plan-hardcoreSkill
- execute-planSkill
- git-commitSkill
- git-pushSkill
- guide-ingestSkill
- ilya-sutskeverSkill
- markdownSkill
- phase-auditSkill
- research-debuggingSkill
- write-notesSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.