persona-query
Skill2 starsChanged 50 days agoMITView on GitHub
Answer "would this persona…?" with three cited scenarios (most likely / least likely / edge case) and confidence Levels. Use on /persona-query — grounded answers without roleplay.
In this repository31 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- affinitySkill
- ai-personaSkill
- analytics-syncSkill
- backlogSkill
- catalogSkill
- cold-startSkill
- contradictionsSkill
- demo-dataSkill
- dovetail-syncSkill
- exportSkill
- extract-findingsSkill
- feature-panelSkill
- forget-participantSkill
- graph-lintSkill
- interview-guideSkill
- journey-mapSkill
- opportunity-treeSkill
- persona-avatarSkill
- persona-talkSkill
- persona-voiceSkill
- persona-workshopSkill
- prdSkill
- prioritizeSkill
- researcherSkill
- retentionSkill
- source-syncSkill
- subject-accessSkill
- welcomeSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
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 22
Before answering, read the Persona file and the Signals/ / Evidence/ files it links to — the grounding lives in those, not in the Persona …
- Distinctive: line 44
Ground every claim in the graph: No scenario from generic user psychology — only from this persona's linked Signals/Evidence.
- Distinctive: line 45
Scenarios must be meaningfully different: Not three versions of "she might use it."
- Very specific: line 46
Cite the source file: Every scenario references at least one Signals/…, Evidence/…, or a named Correlation.
- Very specific: line 51
Freshness rule: If the newest supporting Signal ("Interview date") or Evidence (retrieved:) is older than 3 months, append: 🕒 Data older …
- Very specific: line 52
Competitor context is opt-in: Use Competitors/*.md only when the user asks a competitor-related question or passes --competitors.
≈ “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
persona-query — Query Persona Data
Answer research questions about a persona with structured probability analysis. Always returns 3 scenarios: most likely, least likely, and edge case — each grounded in the persona's linked Signals/ and Evidence/.
Trigger
/persona-query Personas/Emma.md
/persona-query ← lists available personas in Personas/Then the user asks a question, e.g.:
- "Would Emma pay for a smart library feature?"
- "How would she react to a 20% price increase?"
- "Would she recommend Spotify to a friend?"
Before answering, read the Persona file and the Signals/ / Evidence/ files it links to — the grounding lives in those, not in the Persona summary alone.
Response format — always 3 scenarios
🟢 Most likely
What the data strongly suggests. Must cite at least one linked Signal, Evidence, or Correlation. One clear answer + brief reasoning (2–4 sentences).
🔴 Least likely
What the data suggests won't happen — and why. Explain the counter-evidence: which Signal/Evidence rules it out. Not "impossible" — just against the grain of what we know.
⚠️ Edge case
The scenario that could happen under specific conditions the data doesn't rule out. Name the condition: "This could happen IF…" This is the "worst case for the product team" — plan for it even if it's not the default.
Rules
- Ground every claim in the graph. No scenario from generic user psychology — only from this persona's linked Signals/Evidence. If there's no data, say so.
- Scenarios must be meaningfully different. Not three versions of "she might use it."
- Cite the source file. Every scenario references at least one
Signals/…,Evidence/…, or a named Correlation. - Edge case ≠ speculation. It must sit on a real tension in the data — a contradiction, an extreme value, a
validatedfinding pushed to its limit. - State confidence with the Level scale (canonical definition:
ai-persona/references/levels.md— L1 assumption → L2 Evidence-only → L3 Signal-only → L4 validated → L5 Signal+Evidence+Correlation). This is analyst-facing output, so the short code is fine here. After the three scenarios, one line:Confidence: High (L5) — Signals Skip track + Suggested songs, Evidence Smart Shuffle complaints and a Correlation converge. Confidence: Low (L3) — single Signal, no Evidence confirmation.
- Freshness rule. If the newest supporting Signal ("Interview date") or Evidence (
retrieved:) is older than 3 months, append:🕒 Data older than 3 months — a refresh round of research is recommended before deciding. - Competitor context is opt-in. Use
Competitors/*.mdonly when the user asks a competitor-related question or passes--competitors; cite the competitor file explicitly.
Example
Question: Would Emma pay extra for a "Listening Modes" feature?
🟢 Most likely — Yes, if bundled into Premium (not a separate charge) Signals Suggested songs + Skip track; Correlation "Context pollution corrupts the core experience". She explicitly wants a simple context switch and her core frustration is algorithm pollution from context-switching. She'd value it — but as part of what she already pays for.
🔴 Least likely — Paying a standalone add-on price Evidence Price sensitivity (43% cancels a subscription seen as too costly). She already feels Premium is overpriced for her use case; an extra charge for something that should "just work" would confirm it. More likely: a manual workaround than paying.
⚠️ Edge case — She churns entirely if price rises before this is fixed Evidence Price sensitivity + Correlation "Product-market misalignment". Frustration with drift + cost sensitivity compound. This could happen IF a competitor offers profile separation first.
Confidence: High — two Signals and one Evidence point the same way.
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.

