cluster-pains
Skill0 starsChanged 4 months agoMITView on GitHub
Fold research briefs into recurring, evidenced pain clusters.
In this repository11 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- analyze-competitionSkill
- analyze-marketSkill
- collect-evidenceSkill
- compose-reportSkill
- contradiction-analysisSkill
- decisionSkill
- discover-opportunitiesSkill
- harvest-leadsSkill
- interview-planSkill
- research-briefSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 30
If a single fix could not address both, they are separate pains however similar the language.
- Distinctive: line 44
Do not name, sketch, or imply what should be built.
- Distinctive: line 51
No sizing, no revenue, no share of market, no "this affects thousands of teams", no willingness to pay.
- Distinctive: line 78
If a single source describes a real, specific, costly problem, it is a cluster.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Cluster pains
Group what these briefs describe into distinct recurring pains — by the workflow problem people actually have, not by the words they used for it.
The question
${question}The research briefs
${briefs}Cluster by the underlying workflow problem
Vocabulary is a trap in both directions. Two briefs both saying "sync" may describe unrelated problems; two briefs sharing no words at all — one about pasting between tabs, one about a nightly export script — are often the same pain.
The test to apply: would one fix plausibly relieve both? If yes, they belong together. If a single fix could not address both, they are separate pains however similar the language. A second test that catches the rest: describe the moment the pain occurs — who is doing what, at which step, with which tools open. Two descriptions that land on the same moment are one cluster.
Also keep separate:
- the pain from the workaround people built for it,
- the pain from the tool they blame for it,
- the symptom from the cause, when the briefs distinguish them. When they do not, cluster on the symptom — that is what was observed — and say the cause is not established.
No solutions. None.
Do not name, sketch, or imply what should be built. Not as an aside, not in a label, not as "which suggests a need for". The next stage cannot infer an honest opportunity from a problem statement that already contains the answer, and a cluster written around a solution quietly stops being a description of what people said.
No market estimation either
No sizing, no revenue, no share of market, no "this affects thousands of teams", no willingness to pay. Nothing about buyers or budgets. Prevalence is a measurement, not an impression: leave it unset unless the briefs actually measured it.
Name it properly
- Label it in the sufferers' words where they have one. Their phrase for the problem is more useful downstream than your category name for it.
- Describe who hits it, when in the workflow they hit it, and what it costs them. A description that would fit any company in any industry describes nothing.
- Segments narrow enough that someone could go and find these people: "ops leads at Series B companies running their own billing reconciliation", not "businesses" or "users".
- Quotes come verbatim from the briefs. Two sources landing on the same phrasing is itself the finding, so keep the repeat.
Counting and severity
source_countis distinct sources. One source restating itself across three briefs counts once. This number is the whole difference between a signal and an anecdote.- Severity is what the pain costs whoever has it, and observed behaviour is the tell: people paying for a workaround, hiring for it, or actively shopping is high; people complaining while changing nothing is low. Do not read severity off the intensity of the language — angry posts are cheap. Leave it unset if the briefs do not support a reading.
A cluster of one is allowed
If a single source describes a real, specific, costly problem, it is a cluster. Record it honestly as one source rather than dropping it: a sharp anecdote is more useful than a vague aggregate, and the next stage can see the count. What is not allowed is dressing it up as a pattern.
Failure modes
- Keyword clustering: grouping on a shared noun instead of a shared problem.
- One enormous cluster that covers everything and therefore means nothing.
- A leftover "miscellaneous" or "other issues" bucket. Either it is a pain or it is not.
- Overlapping clusters that are one pain sliced by phrasing.
- Clusters that restate the question rather than what the briefs found.
- Inventing a segment the briefs never mention, or inflating a count to justify a cluster you find interesting.
- Any solution, product idea, sizing or buyer speculation leaking in.
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.

