intent-ref-research-methods

A reference for choosing ways to learn about users, such as observation, interviews, usability tests, and surveys. It explains what each method can reveal, when to use it, and how many people to include.

In plain words
What is it for?
Use it to plan user research, select a suitable method, prepare interviews or tests, decide on sample sizes, and organize findings. User research means collecting evidence about people’s needs, behavior, and experience with a product.
Why use it?
It helps avoid using a method that cannot answer the question you have or treating a small set of interviews as proof of what everyone thinks. It makes research choices clearer within limits of time and budget.

Cursor rule for Cursor

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/ghaida/intent/intent-ref-research-methods
Clone the repo
git clone --depth 1 https://github.com/ghaida/intent

Made for: Cursor.

Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,452 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00031 $0.03452
Opus 5 $0.00015 $0.01726
Sonnet 5 $0.00006 $0.00690
Haiku 4.5 $0.00003 $0.00345

Measured 2d ago against content hash 0dc8b125f99d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

intent-ref-research-methods scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.cursor/rules/intent-ref-research-methods.mdc · 214 lines

How it starts

The opening of the file, as written. The whole thing — 214 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Research Methods

Method Selection Matrix

Choosing the right research method depends on what you need to learn, how much time and budget you have, and where you are in the design process. There is no universal "best method" — there's the right method for the question you're asking right now.

Generative Methods (What should we build?)

Contextual Inquiry

  • What it is: Observe users in their natural environment while they perform real tasks. Ask questions as they work.
  • When to use: Early exploration. You don't understand the problem space well enough to ask good survey questions yet.
  • Sample size: 4-8 participants per user segment.
  • Time/cost: High. 1-2 hours per session, plus travel. Analysis is intensive.
  • What it reveals: Workarounds, environmental constraints, unspoken needs, the gap between what people say and what they do.
  • Trade-offs: Small samples, not generalizable, observer effect can alter behavior. But the depth of insight is unmatched.

Semi-Structured Interviews

  • What it is: One-on-one conversations guided by a topic framework, not a rigid script. Follow interesting threads.
  • When to use: When you need to understand motivations, mental models, and experiences in depth. Works at any stage.
  • Sample size: 5-8 for pattern identification, 12-20 for saturation (Guest, Bunce & Johnson, 2006).
  • Time/cost: Moderate. 45-60 minutes per session. Analysis takes roughly 3x the interview time.
  • What it reveals: User motivations, pain points, mental models, emotional responses, workarounds.
  • Trade-offs: Self-reported behavior differs from actual behavior. Users are not reliable predictors of their own future actions. But interviews surface the "why" that behavioral data can't.

Diary Studies

  • What it is: Participants record experiences over time (days to weeks), logging entries when specific events occur.
  • When to use: When behavior unfolds over time and can't be observed in a single session. Habit formation, recurring tasks, infrequent events.
  • Sample size: 10-15 participants minimum (high dropout expected — recruit 20-30% more).
  • Time/cost: Study runs 1-4 weeks. Setup is moderate; analysis is substantial.
  • What it reveals: Temporal patterns, context shifts, emotional changes over time, frequency and triggers of behavior.
  • Trade-offs: High participant burden, significant dropout, entries are self-reported and often incomplete. But nothing else captures real behavior over time.

Read the full file on GitHub · 214 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 214 lines · 3,452 tokens per session scan A 0dc8b125f99d

Subscribe to this mod's changes

intent-ref-research-methods is a cursor rule published in the GitHub repository ghaida/intent (139 stars, last pushed 1mo ago), licensed CC0-1.0. It adds 31 tokens to every session and 3,452 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.