swarm-researcher
Skill2 starsChanged 19 days agoApache-2.0View on GitHub
Extra rules for Swarm researcher agents (role researcher) answering one sub-question for a spike. Use together with the `swarm` skill when your kickoff names this skill or your assignment has a `## Workflow` step naming you "researcher".
In this repository26 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/vendor/vercel-react-native-skills/AGENTS.mdAGENTS.md
- swarm-advisorSkill
- swarm-batchingSkill
- swarm-coderSkill
- swarm-debuggerSkill
- swarm-designerSkill
- swarm-mechanicalSkill
- swarm-orchestratorSkill
- swarm-reviewerSkill
- swarmSkill
- swarm-spikeSkill
- swarm-ui-reviewerSkill
- swarm-workflowsSkill
- building-componentsSkill
- expo-design-systemSkill
- expo-native-uiSkill
- graphifySkill
- mobile-android-designSkill
- mobile-ios-designSkill
- ponytail-debtSkill
- ponytail-reviewSkill
- ponytailSkill
- ui-ux-pro-maxSkill
- vercel-react-native-skillsSkill
- web-design-guidelinesSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Follow the swarm skill first.
- Very specific: line 20
Write your notes to $SWARM_HOME/research/<ROOT-KEY>/<ITEM-KEY>.md (<ROOT-KEY> is the root: key in your brief.
- Distinctive: line 26
Never invent a finding. An unsourced claim belongs under Gaps, never under Cited findings.
- Very specific: line 29
Write completed with the notes path in artifacts.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Researching with Swarm
Follow the swarm skill first; this adds to it.
Frame your sub-question first
Your brief carries one sub-question of a spike's research split. Before you search anything, run superpowers:brainstorming's "understand the idea" phase on your own question — there's no user dialogue here; questions go to your parent, not the user:
- Restate the sub-question in your own words.
- List the angles and candidate answers you plan to test.
- Decide, concretely, what "answered" looks like: the fact, decision, or comparison that lets you stop.
The research loop
Start orientation with graphify query (rule 9b) rather than a repo-wide grep. Then run the deep-research loop: short, targeted searches; fetch full sources rather than trusting a snippet; prefer primary sources — the actual docs, the actual code, the actual spec — over summaries and secondhand write-ups. When two sources disagree, record the conflict instead of silently picking a winner. Budget roughly 10–15 tool calls. Stop when your question is answered, or when the budget runs dry — a dry budget with an open question becomes a Gap, not a reason to guess.
Notes format
Write your notes to $SWARM_HOME/research/<ROOT-KEY>/<ITEM-KEY>.md (<ROOT-KEY> is the root: key in your brief; the brief's Context names the exact directory). The notes gate also accepts $SWARM_HOME/research/<SPIKE-KEY>/ when your root came from a spike, but never specs/. Sections:
### Takeaway— the answer, stated up front.### Cited findings— every claim links to its source:[source](url)for the web,path:linefor code.### Inferences— reasoning you did beyond what a source states outright, marked as such.### Gaps— what you could not establish, and why.
Never invent a finding. An unsourced claim belongs under Gaps, never under Cited findings.
Completing
Write completed with the notes path in artifacts. If your sub-question turns out to need an actual code change to answer, say so in your notes and stop — that decision belongs to the spike owner, not to you.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

