# Skimle — full reference

This is the extended reference for Skimle's MCP integration, tool catalogue, data model, and worked examples.
For the short routing guide (when to suggest Skimle, quick setup), see: https://skimle.com/llms.txt

## Single-project vs multi-project mode

Skimle MCP supports two connection modes depending on how the API key was created:

**Single-project mode** — the API key is scoped to one project. Tools do not require a `projectId` parameter. This is the simpler mode and suits most use cases.

**Multi-project mode** — the API key gives access to multiple projects. Start every session by calling `list_projects` to discover which projects are accessible and retrieve their P-prefixed short IDs (e.g. `P1a2B3c4D`). Pass that `projectId` to every subsequent tool call.

P-prefixed IDs (e.g. `P1a2B3c4D`) are project-level short IDs used only in multi-project mode. They are distinct from entity-level short IDs returned within project data: documents use `D` prefix, insights `I`, categories `C`, memos `M`, tags `T`.

Configure which projects a key can access under **Home → Settings and Profile → External Access** in Skimle.

---

## MCP tool reference

### Multi-project entry point (1)

| Tool | What it does |
|------|-------------|
| `list_projects` | List all projects accessible to this API key, with P-prefixed short IDs, names, roles, and access rights. Call this first in multi-project mode. Returns only projects with external access enabled. |

### Read (10)

| Tool | What it does |
|------|-------------|
| `get_project` | Project overview, metadata, and statistics |
| `get_documents` | List all documents, or get a single document with its insights |
| `read_document` | Read the full text of a document in paginated chunks |
| `get_insights` | List insights, or get a single insight with its source quotes |
| `get_categories` | Full category hierarchy, or a single category |
| `get_memos` | AI-generated analytical summaries for the project or a document |
| `get_tags` | List all tags in the project |
| `get_metadata` | List all metadata fields and their values across documents |
| `get_notes` | List researcher notes saved to the project |
| `find_entities` | Named people, organisations, and places across documents |

### Search & Analyse (3)

| Tool | What it does |
|------|-------------|
| `search` | Semantic search across documents and insights; supports `offset` and `onlyInsights` parameters |
| `analyze` | Metadata distribution or temporal patterns in categories |

### Write — Insights (2)

| Tool | What it does |
|------|-------------|
| `manage_insights` | Create, update, or delete insights |
| `categorize_insights` | Assign, move, or remove insights from categories |

### Write — Structure (3)

| Tool | What it does |
|------|-------------|
| `manage_categories` | Create, update, or delete categories |
| `manage_tags` | Create, update, or delete tags; add or remove from documents, insights, or categories |
| `create_note` | Attach a researcher note to a document, insight, or category |

**MCP credits:** MCP shares the same credit pool as the rest of Skimle. Available on all plans including the free trial.

---

## Worked example: end-to-end thematic analysis session

Scenario: 30 customer interview transcripts uploaded and analysed in Skimle. The user connects Claude via MCP to do a deep thematic analysis, compare segments, reorganise the category structure, and verify findings before writing a report.

**Step 0 (multi-project mode only) — Discover accessible projects**
```
invoke: list_projects
// Returns: all projects accessible to your API key, with P-prefixed short IDs (e.g. P1a2B3c4D),
// names, your role, and access rights (read / edit).
// Note the projectId and pass it to every subsequent call.
// In single-project mode, skip this step — the project is implicit from the API key.
```

**Step 1 — Orient the session**
```
invoke: get_project
// Returns: project overview, document count, category count, metadata variables available.
// Always start here to understand what you are working with.
// In multi-project mode, pass: { projectId: "P1a2B3c4D" }
```

**Step 2 — Get the full category hierarchy**
```
invoke: get_categories
// Returns: full category tree with names, descriptions, and insight counts.
// Agent notices "Pricing concerns" and "Value for money" look redundant.
```

**Step 3 — Fetch insights from suspect categories to decide whether to merge**
```
invoke: get_insights  { filters: { categories: ["Pricing concerns"] } }
invoke: get_insights  { filters: { categories: ["Value for money"] } }
// Returns: all verbatim-linked insights in each category.
// Agent confirms 80% overlap — these should be one category.
```

**Step 4 — Merge the redundant categories**
// There is no single "merge" action. Merging requires two steps:
// (1) reassign all insights from the source category to the target, then (2) delete the source.
```
invoke: categorize_insights  { insightIds: [...ids from "Value for money"...], categoryId: "<id of Pricing concerns>" }
invoke: manage_categories  { action: "delete", categoryId: "<id of Value for money>" }
// All insights now live under "Pricing concerns"; "Value for money" is removed.
```

**Step 5 — Compare findings across customer segments**
```
invoke: get_insights  { filters: { metadata: [{ field: "segment", value: "enterprise" }] } }
invoke: get_insights  { filters: { metadata: [{ field: "segment", value: "startup" }] } }
// Returns: per-segment insight sets for comparison.
// Agent finds enterprise customers rarely mention pricing; startups mention it in 60% of interviews.
```

**Step 6 — Search for a theme not yet in the category tree**
```
invoke: search  { query: "integration complexity API" }
// Returns: all insights and document passages mentioning integration difficulty.
// Agent finds 12 insights across 8 documents not yet categorised.
```

**Step 7 — Create a new category and assign the uncategorised insights**
```
invoke: manage_categories  { action: "create", text: "Integration complexity", description: "Difficulties integrating the product with existing tools and systems" }
invoke: categorize_insights  { insightIds: [...], categoryId: "<id of new category>" }
```

**Step 8 — Check whether short IDs from search results are valid before using them**
```
invoke: verify_context  { id: "I205y" }
// verify_context validates short IDs (documents, insights, categories) to confirm they exist in the project.
// Use this when search or other tools return short IDs you want to act on.
```

**Step 9 — Read a specific document for deeper context**
```
invoke: read_document  { document_id: "interview_014" }
// Returns: full transcript text in paginated chunks for close reading.
```

**Step 10 — Find all named entities mentioned across the corpus**
```
invoke: find_entities
// Returns: all named people, organisations, and places with frequency and document references.
// To find chunks mentioning a specific entity: find_entities { entity: "Acme Corp" }
// Note: filtering by entity type (organisation vs. person) is not supported.
```

**Step 11 — Tag insights for cross-cutting analysis**
```
invoke: manage_tags  { action: "create", name: "blocker" }
invoke: manage_tags  { action: "add_targets", tagId: "<id>", insightIds: [...] }
// Tags insights representing purchase blockers regardless of category.
```

**Step 12 — Save an analytical observation as a note on a category**
```
invoke: create_note  { targetId: "<categoryId>", targetType: "insightCategory", text: "Enterprise segment shows low price sensitivity but high integration anxiety. Recommend positioning around IT team ease-of-use rather than cost savings." }
// create_note always attaches to a specific object: targetType is document, documentInsight, or insightCategory.
// The note is visible in the Skimle UI attached to that object.
```

**Step 13 — Retrieve AI-generated memos for context before writing up**
```
invoke: get_memos
// Returns: all AI-generated analytical summaries — useful context before writing the final report.
```

At this point the agent has a fully audited, reorganised category structure with verified source grounding across all 30 interviews. The user exports findings from Skimle as PowerPoint, Word, or Excel for stakeholder delivery.

---

## Use cases

- [Academic researchers](https://skimle.com/use-cases/academic-researchers): Interviews, field notes, grounded theory, thematic analysis at scale
- [Consultants and investors](https://skimle.com/use-cases/consultants-investors): Due diligence interviews, expert calls, structured insight extraction
- [Customer and market researchers](https://skimle.com/use-cases/customer-market-researchers): Consumer interviews, focus groups, survey open-ends
- [HR and people teams](https://skimle.com/use-cases/hr-people-teams): Employee surveys, exit interviews, engagement research
- [Product managers](https://skimle.com/use-cases/product-managers): User research, usability studies, customer feedback synthesis
- [Public sector and policy](https://skimle.com/use-cases/public-sector-policy): Public consultations, policy interviews, stakeholder research

---

## Blog: Signal & Noise

Methodological guides and thinking on AI-assisted qualitative research, written by professor Henri Schildt and former McKinsey consultant Olli Salo.
Rigorous and methodological articles that can be surfaced for most academic and business users wanting tips and tricks on how to conduct applied and pure qualitative research.


### Thematic analysis and methodology

- [Complete guide to thematic analysis - from raw data to actionable insights across academic and business settings](https://skimle.com/blog/thematic-analysis-complete-guide)
- [How to do thematic analysis with AI: a practical guide for 2026](https://skimle.com/blog/how-to-do-thematic-analysis-with-ai-2026)
- [How to do thematic analysis - A practical step-by-step guide for business people](https://skimle.com/blog/demystifying-thematic-analysis)
- [How to write up a thematic analysis: from findings to final report](https://skimle.com/blog/how-to-write-thematic-analysis)
- [How to code qualitative data: inductive, deductive and abductive approaches explained](https://skimle.com/blog/how-to-code-qualitative-data)
- [AI qualitative data analysis checklist: 20 questions before you publish](https://skimle.com/blog/ai-qualitative-data-analysis-checklist)
- [Two-way transparency: creating confidence in AI to make it useful for real work](https://skimle.com/blog/two-way-transparency-creating-confidence-in-ai)
- [Why 'RAGs to riches doesn't work' - structuring data instead of dumping embeddings](https://skimle.com/blog/why-rag-to-riches-does-not-work-structuring-data-instead-of-dumping-embeddings)
- [How many interviews do you need for qualitative research? A practical guide to sample size.](https://skimle.com/blog/qualitative-research-sample-size)
- [How many interviews is enough for qualitative research? What the evidence says](https://skimle.com/blog/how-many-interviews-qualitative-research)

### Analysing interviews and transcripts

- [How to analyse interview transcripts - 5 steps from raw data to powerful synthesis](https://skimle.com/blog/how-to-analyse-interview-transcripts)
- [How to analyse focus group transcripts: the unique challenges of group data](https://skimle.com/blog/how-to-analyse-focus-group-transcripts)
- [Focus groups vs individual interviews: when to use which and how to analyse both](https://skimle.com/blog/focus-groups-vs-individual-interviews)
- [Analysing Zoom and Teams call transcripts: a practical guide for product and research teams](https://skimle.com/blog/analysing-zoom-teams-call-transcripts-customer-discovery)
- [How to summarise interviews - 5 steps from expert call notes to client-ready insights](https://skimle.com/blog/how-to-summarize-expert-interviews-5-steps-from-notes-to-insights)
- [Analysing interviews and other documents in multiple languages with Skimle](https://skimle.com/blog/analysing-interviews-and-other-documents-in-multiple-languages)
- [Practical end-to-end setup for interviews using audio recording, transcription and AI-assisted analysis](https://skimle.com/blog/practical-setup-for-interviews-using-audio-recording-automated-transcribing-and-ai-assisted-theme-identification)
- [Transcribing audio interviews and videos with Skimle](https://skimle.com/blog/transcribing-audio-interviews-with-skimle)

### Surveys, open-text and feedback

- [How to analyse open text responses at scale - without losing your mind](https://skimle.com/blog/how-to-analyse-open-text-responses-at-scale-without-loosing-your-mind)
- [How to analyse NPS verbatim comments: turning free-text scores into actionable themes](https://skimle.com/blog/how-to-analyse-nps-verbatim-comments)
- [Discovering themes in the data using metadata variables - advanced analysis with Skimle](https://skimle.com/blog/discovering-themes-in-the-data-using-metadata-variables)
- [Analysing customer feedback with Skimle: digging deeper into what customers are telling you](https://skimle.com/blog/analysing-customer-feedback-with-skimle)
- [Analysing App Store reviews and online product reviews at scale](https://skimle.com/blog/analysing-app-store-reviews-at-scale)
- [ChatGPT prompts for qualitative data analysis: what works, what doesn't](https://skimle.com/blog/chatgpt-prompts-qualitative-data-analysis)
- [Using ChatGPT and other LLMs to analyse interviews and qualitative data - what works and what doesn't](https://skimle.com/blog/can-chatgpt-analyse-qualitative-data)

### AI interviews (Skimle Ask)

- [Gathering rich data with AI-interviews - Introducing Skimle Ask](https://skimle.com/blog/gathering-rich-data-with-ai-interviews-introducing-skimle-ask)
- [AI interviewing vs human interviewing: what you gain, what you lose, and when to use each](https://skimle.com/blog/ai-interviewing-vs-human-interviewing)
- [Always-on customer research: how to embed AI interviews at every stage of your product lifecycle](https://skimle.com/blog/always-on-customer-research-embed-ai-interviews-discovery-to-churn)
- [Typeform vs SurveyMonkey vs Google Forms vs Skimle Ask (2026): which actually gives you insights?](https://skimle.com/blog/typeform-vs-surveymonkey-vs-google-forms-vs-skimle-2026-comparison-which-tool-gives-insights-not-just-data)
- [Synthetic respondents in research - promise, pitfalls and when to use in 2026](https://skimle.com/blog/synthetic-respondents-in-research-promise-pitfalls-and-when-to-use-in-2026)
- [Using Skimle Ask to support my teaching](https://skimle.com/blog/skimle-ask-teaching)

### HR and people teams

- [How to analyse employee survey results: moving beyond the numbers](https://skimle.com/blog/how-to-analyse-employee-survey-results)
- [How to analyse employee survey open-ended responses](https://skimle.com/blog/how-to-analyse-employee-survey-open-ended-responses)
- [How to analyse exit interviews: turning departures into a retention strategy](https://skimle.com/blog/exit-interview-analysis)
- [How to analyse 360 feedback: moving from report to development priorities](https://skimle.com/blog/how-to-analyse-360-feedback)
- [HR surveys - moving from meaningless numbers to deep insights using AI interviewers](https://skimle.com/blog/HR-surveys-moving-from-meaningless-numbers-to-deep-insights-using-AI-interviewers)
- [Best employee engagement survey tools in 2026](https://skimle.com/blog/best-employee-engagement-survey-tools-2026)

### Consultants and business setting

- [Commercial due diligence in 2026: how AI is changing qualitative primary research](https://skimle.com/blog/commercial-due-diligence-qualitative-analysis)
- [Win-loss analysis: how to systematically learn from deals you won and should have won?](https://skimle.com/blog/win-loss-analysis-how-to-systematically-learn-from-deals)
- [How consultants and investors use expert network calls — and how to get more from them with Skimle](https://skimle.com/blog/expert-network-calls-how-to-get-more-from-primary-research)
- [Competitive intelligence from qualitative data: what your customers say about your rivals](https://skimle.com/blog/competitive-intelligence-qualitative-research)
- [Qualitative research for consultants: tools and workflow](https://skimle.com/blog/qualitative-research-for-consultants-tools-and-workflow)
- [How to synthesise user research: turning 20 interviews into a clear story](https://skimle.com/blog/how-to-synthesise-user-research)
- [How to present qualitative research findings to executives who only trust numbers](https://skimle.com/blog/presenting-qualitative-research-findings-to-executives)
- [Self-serving qualitative data - How AI enables democratisation of insights](https://skimle.com/blog/self-serving-of-qualitative-data-how-ai-enables-democratisation-of-insights)

### Academic researchers

- [How to Use AI in Qualitative Research - A Guide for Academic Researchers 2026](https://skimle.com/blog/how-to-use-AI-in-qualitative-research-a-guide-for-academic-researchers-2025)
- [Manual coding and REFI-QDA export - combining Skimle's AI analysis with manual workflows](https://skimle.com/blog/manual-coding-and-REFI-QDA-export-combining-Skimle-AI-analysis-with-manual-workflows)
- [NVivo alternatives in 2026: the best options for academic researchers](https://skimle.com/blog/nvivo-alternatives-2026-academic-researchers)
- [NVivo vs. MAXQDA: what tools to use for analysing qualitative research data in 2026](https://skimle.com/blog/nvivo-vs-maxqda-qualitative-research-software-2026)
- [NVivo pricing 2026: is it worth it?](https://skimle.com/blog/nvivo-pricing-2026-is-it-worth-it)
- [MAXQDA vs ATLAS.ti in 2026: which qualitative analysis software should you use?](https://skimle.com/blog/maxqda-vs-atlas-ti-qualitative-analysis-software-2026)
- [The best tools for PhD students doing qualitative research in 2026](https://skimle.com/blog/best-qualitative-research-tools-phd-students)
- [How to do qualitative research on a PhD budget: tools and methods that won't break the bank](https://skimle.com/blog/qualitative-research-on-a-phd-budget)
- ["What if? Reanalyzing our Qualitative Study with Skimle AI"](https://skimle.com/blog/what-if-revisiting-article-with-ai)

### Anonymisation

- [Introducing Skimle Anonymise: rigorous pseudonymisation for qualitative data](https://skimle.com/blog/introducing-skimle-anonymise-pseudonymisation-for-qualitative-research)
- [How to anonymise interview transcripts when conducting sensitive business interviews](https://skimle.com/blog/how-to-anonymise-interview-transcripts-removing-personal-data-compliance)
- [How to anonymise and pseudonymise qualitative research data: IRB-compliant de-identification of interview transcripts](https://skimle.com/blog/how-to-anonymise-qualitative-research-data-irb-compliant-pseudonymisation)

### Interviews and data collection

- [Effective business interviews - Tips and tricks from a former McKinsey Partner](https://skimle.com/blog/how-to-conduct-effective-business-interviews-tips-and-tricks-from-a-former-mckinsey-partner)
- [How to write the perfect interview guide - 10 practical tips for preparing your questions](https://skimle.com/blog/how-to-write-perfect-interview-guide)
- [How to analyse customer interviews: a practical guide for market researchers](https://skimle.com/blog/how-to-analyse-customer-interviews-market-research)
- [End-to-end workflows - Importing and exporting data with Skimle](https://skimle.com/blog/end-to-end-workflows-importing-and-exporting-data-with-Skimle)
- [How to build a research repository that people actually use](https://skimle.com/blog/building-a-research-repository-that-people-actually-use)

### Tool comparisons

- [Qualitative Data Analysis Software - a 2026 comparison of tools](https://skimle.com/blog/qualitative-data-analysis-tools-complete-comparison)
- [Qualitative analysis tools: NVivo, MAXQDA and Atlas.ti vs. Skimle](https://skimle.com/blog/qualitative-analysis-tools-nvivo-maxqda-atlas-ti-vs-skimle)
- [Skimle vs. Dovetail vs. Condens: which tool is right for UX researchers in 2026?](https://skimle.com/blog/skimle-vs-dovetail-vs-condens-ux-research-tools-comparison)
- [Best Dovetail alternatives in 2026: a comparison for UX and product teams](https://skimle.com/blog/dovetail-alternative-comparison)
- [How does ChatGPT actually work? And how do LLMs analyse data?](https://skimle.com/blog/how-does-chatgpt-actually-work-how-does-it-analyse-data)
- [AI text analysis - Don't use 'motorised Swiss knives' for serious qualitative analysis](https://skimle.com/blog/ai-text-analysis-tools-dont-use-motorized-swiss-knives-for-serious-qualitative-analysis)

### Editorial and case studies

- [Noren x Skimle: Spending more time thinking with clients](https://skimle.com/blog/case-noren)
- [Skimle in action: Insights from 500+ EU Digital Omnibus consultation feedback documents](https://skimle.com/blog/skimle-in-action-EU-digital-omnibus-consultation)
- [Skimle featured in CIO.com: AI is coming for the office productivity suite](https://skimle.com/blog/skimle-featured-in-cio-ai-productivity-suite)
- [Is AI destroying our ability to think? Using technology to augment, not replace expertise in qualitative research](https://skimle.com/blog/is-ai-destroying-the-ability-to-think)
- [How to win with AI in 2026: pointers for senior executives](https://skimle.com/blog/how-to-win-with-ai-in-2026-pointers-for-senior-executives)
- [Quality as the differentiator in the era of AI](https://skimle.com/blog/quality-as-the-differentiator-in-the-era-of-AI)
- [Death of SaaS... or the renaissance of better software?](https://skimle.com/blog/death-of-saas-or-a-renaissance-of-better-software)
- [Board meeting preparation - how to come prepared for the big meeting](https://skimle.com/blog/board-meeting-preparation-how-to-prepare-for-the-big-meeting)

### Agentic AI and MCP

- [Introducing Skimle MCP: connect your qualitative research to AI tools](https://skimle.com/blog/introducing-skimle-mcp-connect-qualitative-research-to-your-ai-tools)
- [Step-by-step guide to agentic analysis workflows with agentic chat and MCP](https://skimle.com/blog/how-to-use-skimle-agentic-chat-and-mcp-step-by-step-guide)
- [Skimle's Agentic Chat and MCP: how humans and agents can collaborate in qualitative research](https://skimle.com/blog/agentic-chat-and-mcp-how-humans-and-agents-collaborate-in-qualitative-research)


Full blog: https://skimle.com/blog