Search Jupiter's user research library on Google Drive and surface relevant findings for a topic, question, or PRD. Scans both research folders, identifies relevant studies, reads their reports, and outputs findings either as a quick reference or as a formatted PRD evidence block. Use when writing a PRD, validating a problem statement, or checking what Jupiter already knows about a topic before starting new research.
From pm-market-researchnpx claudepluginhub jupitermoney/pm-superic-skills --plugin pm-market-researchThis skill uses the workspace's default tool permissions.
Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.
Migrates code, prompts, and API calls from Claude Sonnet 4.0/4.5 or Opus 4.1 to Opus 4.5, updating model strings on Anthropic, AWS, GCP, Azure platforms.
Details PluginEval's skill quality evaluation: 3 layers (static, LLM judge), 10 dimensions, rubrics, formulas, anti-patterns, badges. Use to interpret scores, improve triggering, calibrate thresholds.
Google Workspace MCP must be connected before this skill can run.
Silently attempt to list files from the primary research folder. If the call fails or returns an auth error, stop immediately and tell the user:
"This skill requires Google Workspace MCP to access Jupiter's research library. Connect the Google Workspace MCP integration and restart. Without it, the skill cannot run."
Do not proceed past this check if MCP is unavailable.
Two research folders are searched on every run. Both are hardcoded — the PM never needs to find or share folder links.
| Library | Folder ID | Style |
|---|---|---|
| UX Research @Jupiter | 18j1txWhy5LAXhh3MMJpN79eho36e_jM8 | Structured — numbered subfolders, reports buried inside. Native Google Docs and Slides. Studies from 2025–2026. |
| Jupiter Research | 1J0ExN3pBHEpHJUM9Gb9z5xIXjpnZrOwT | Flat — files sit directly in the study folder. Office format files (.docx insights docs, .pptx presentations). Studies from 2023–2025. |
If either folder returns a "File not found" or permission error, note which one is inaccessible and continue with the other. Do not fail the whole skill. Tell the user:
"Note: [folder name] is not accessible with the current Google account. Results will only cover the accessible library. To fix this, ensure the folder is shared with abhishek.choudhari@jupiter.money."
Accept $ARGUMENTS directly as the research topic or question.
If $ARGUMENTS is empty, ask one question:
"What are you researching? A topic (e.g. 'SA Pro churn', 'onboarding drop-off', 'credit card benefits') or a specific question works. Also: is this for a PRD, or a quick reference lookup?"
The answer to the second part selects the output mode in Phase 4. Default to Quick Lookup if not specified.
List the top-level contents of both research folders using list_files. Do not read any file content yet.
For each study folder found, score its name against the query on a simple 3-point scale:
Exclude non-study folders: Participant CRM, Process Docs, How-To: Research, UXR Sessions, Toolkits, Product Profiles, Research Processes & SOP, Previous Research Materials (Saloni), UXR Hiring.
Tell the user what was found before reading anything:
"Found [N] studies across [X] librar(y/ies). [N] look relevant to '[topic]':
- [Study name] ([date]) — Strong match
- [Study name] ([date]) — Possible match
Reading their reports now…"
Token guard: If more than 5 studies match, list them all and ask:
"Found [N] relevant studies — reading all of them will use significant context. Want me to read all [N], or narrow to the top 3 most relevant?"
Wait for confirmation before proceeding.
For each matched study, navigate to the report using the structure detection rules below. Then export and extract.
Structured folders (Researcher 1 — numbered subfolders):
After listing the study folder contents, if you see numbered subfolders (e.g. 1. Research Plan, 2. Raw Data, 3. Reports):
Reports, Final Reports, 3. Reports, or 5. Final Reportstext/plainFlat folders (Researcher 2 — files at root):
If files appear directly in the study folder with no numbered subfolders:
.docx (Word) for text-based reports, .pptx (PowerPoint) for presentation reports.docx over .pptx when both exist — text extraction is cleanerexport_file with mimeType: text/plain — this works for both Office and native Google formats.xlsx, .csv, and .pdf files — these are raw data, not insights reportsFrom each report, extract:
Do not read raw data folders (Survey Responses, Raw Data, Session Recordings). Reports and Final Reports only.
If a report is a Google Slides presentation: export as text/plain and extract slide titles and body text. The structure is less clean than a Doc — focus on slide titles that name findings and any data callouts in the body.
Use this when the PM wants a fast reference or is exploring what exists.
## Research findings: [topic]
Searched [N] studies across Jupiter's research library. [N] relevant.
---
**[Study Name]** | [Date] | [Method] | N=[n]
- [Finding — concrete, with number or data point if available]
- [Finding]
- [Finding]
> "[Direct quote if available]"
Source: [Drive link]
---
**[Next study]** | ...
Use this when the PM says this is for a PRD.
## Research Evidence — [topic]
> Use this section in your PRD's Problem Statement or User Insights.
> Each block is one study. Edit the implication line to fit your specific initiative.
---
**[Study Name]** ([Date] | [Method] | N=[n])
**Key finding:** [1–2 sentence finding, grounded in data]
**Supporting data:** [specific stat, percentage, or quote]
**Implication:** [1 sentence on what this means for the PRD — keep it factual, not prescriptive]
[Drive link]
---
For the Implication line: write it as "This suggests [product area] has [specific gap/opportunity]" — not as a recommendation. Leave the PM to draw the conclusion.
After output, offer:
/synthesize-research across these findings to identify themes?"/create-prd [topic] and reference this evidence."