Cookiy AI User Research Skill
End-to-end user research for AI agents — plan, run, and synthesize interviews and surveys.
Evidence shows: README requires users to manually enable network access and set domain allowlists, reflecting least privilege; but no user confirmation mechanism or detailed data flow transparency is provided. Sensitive data handling: README mentions research data syncing with Cookiy platform but does not specify data protection measures. Dependency security: no dependency list or security audit provided. External effects: skill calls external APIs and may recruit real participants, but impact scope is not described. Rollback: no rollback mechanism provided. Source attribution: MIT license and copyright notice exist, but publisher is unverified. Deductions: lack of user confirmation, data flow transparency, and sensitive data handling details; dependency security not mentioned; rollback mechanism missing.
Evidence shows: README's described features are consistent with installation instructions, but no error handling or failure message examples are provided. Dependency availability: relies on external services (Cookiy AI) and network access, but service availability guarantees are not stated. Deductions: failure messages not mentioned, dependency availability not explicit.
Evidence shows: README identifies target audiences (Claude, Codex, Cursor, OpenClaw, etc.) and multiple use cases (interviews, surveys, report synthesis). Capability boundaries are described via feature table but not explicit limitations. Trigger precision: explicit command and semantic matching mentioned, but precision not detailed. Environment fit: multiple installation methods provided, but not all environments covered. Deductions: capability boundaries and trigger precision not detailed enough.
Evidence shows: README has clear structure with sections for installation, usage, examples. Installation notes are detailed, but no FAQ provided. Naming stability: skill name and command are clear, but version compatibility not stated. Known limitations not explicitly listed. License is MIT, but versioning/changelog not provided. Maintenance responsibility: maintainers or update strategy not clearly stated. Deductions: missing FAQ, known limitations, versioning/changelog, and maintenance responsibility.
Evidence shows: Outputs are research reports, interview guides, etc., which are practical. Marginal value: provides automated research capabilities, but no comparison with other tools. Cost-benefit: no pricing or cost information provided. Deductions: cost-benefit not addressed.
Evidence shows: Claims in README (e.g., 900K+ views) lack verifiable sources. Cross-source corroboration: no third-party verification provided. Fact-inference separation: README mixes product description with user testimonials but does not clearly distinguish. Deductions: claims lack traceability, cross-source corroboration insufficient, fact-inference separation not explicit.
- This skill requires network access and may send data to external services (Cookiy AI); ensure data privacy and compliance.
- Publisher identity is unverified; use with caution and assess risks yourself.
- User testimonials and view counts in README lack verifiable sources; do not rely on them for reliability.
What does this agent do, and when should you use it?
The Cookiy AI User Research Skill is an MIT-licensed, open-source agent skill for AI agents including Claude Code, CodeX, Cursor, and OpenClaw. It enables agents to plan, run, and synthesize qualitative (interviews) and quantitative (surveys) user research without leaving the conversation. The skill includes capabilities for research planning, report synthesis, AI-moderated interviews, and multi-language surveys. It is installed as a plugin for Claude desktop apps, a CLI plugin for Claude Code, or via npx for other agents. Usage requires a Cookiy AI account and network access to the Cookiy AI API at s-api.cookiy.ai.
The skill lets an AI agent conduct user research workflows. It generates research plans, screening questionnaires, and interview guides based on user goals. It synthesizes interview transcripts into structured reports with codebooks, personas, and prioritized findings. The skill integrates with the Cookiy AI platform to run AI-moderated interviews and surveys, recruiting real or synthetic participants via the API endpoint s-api.cookiy.ai. For Claude desktop apps, it installs as an uploaded ZIP skill; for Claude Code, via plugin commands; for other agents, it provides a CLI via npx. The command '/user-research-cookiy' explicitly invokes the skill, otherwise the agent loads it automatically based on semantic matching.
- A product manager wants to understand why users churn after onboarding; the skill designs a research plan, screener, and interview guide.
- A UX researcher has 20 interview transcripts and needs a structured report with themes and personas.
- A developer wants to study how developers choose CI/CD tools; the skill runs AI-moderated interviews via Cookiy AI.
- A product team wants to survey 200 users about feature satisfaction; the skill designs the survey, recruits respondents, and collects results.
- An indie developer wants to gather quick user feedback before a design decision without leaving their coding environment.
What are this agent's strengths and limitations?
- Open source with MIT license, allowing free use and modification.
- Supports multiple AI agent platforms (Claude Code, CodeX, Cursor, OpenClaw), offering flexibility.
- Provides end-to-end research capabilities from planning to synthesis, including real or synthetic participants.
- Supports multi-language surveys with conditional logic and respondent recruitment.
- Has significant creator traction (900K+ views and 5,000+ comments), indicating real-world usage and community validation.
- Platform-specific: requires a Cookiy AI account and network access to its API (s-api.cookiy.ai), meaning data leaves your local environment.
- Installation varies by agent and can be configuration-heavy, especially regarding network whitelist settings.
- Core functionality for surveys and interviews heavily depends on Cookiy AI infrastructure; if the service is unavailable or discontinued, core features break.
How do you install or deploy this agent?
For Claude Code terminal: run '/plugin marketplace add cookiy-ai/user-research-skill', then '/plugin install user-research@cookiy-ai', then '/reload-plugins'. For Claude desktop apps: download the ZIP from the GitHub release page, then upload via Customize > Skills > Create Skill > Upload a skill. For Codex, Cursor, or other agents: run 'npx cookiy-ai'. After installation, for Claude desktop apps, you must enable network access in settings and set the domain allowlist to 'All domains' or add 's-api.cookiy.ai'.
How do you use this agent?
After installation, type '/user-research-cookiy' to invoke the skill explicitly, or simply describe your research goal and the agent will load the skill automatically based on semantic matching. For example, say 'I want to understand why users churn after onboarding' or 'Here are 20 interview transcripts, give me a report'. For features requiring Cookiy AI end-to-end studies, ensure you have a Cookiy AI account and log in at cookiy.ai to sync studies, participants, and results.