Research & Evaluation Corpus Setup
$39
Included with membership · or buy it individually
Research & Evaluation Corpus Setup
Scaffold a navigable, version-controlled corpus for any qualitative research or evaluation project. Locks in scope and citation decisions upfront (charter), ingests source documents into a markdown layer, runs a parallel KII/FGD transcript pipeline, and maintains three living navigation files (INDEX, SECTION-MAP, PROJECT-CONTEXT) so the corpus stays traversable as it grows. Pairs with research-synthesis-pipeline, which consumes the corpus this skill produces.
Installs into Claude Code, opencode, Claude.ai, ChatGPT, Gemini. Members get the full library.
It triggers on
- start a new evaluation
- set up a research project
- new evaluation project
- ingest source documents
- convert docx to markdown
- convert PDF to markdown
- OCR a scanned PDF
- transcript pipeline
- anonymise KIIs
- build a corpus
- project charter
- set up CLAUDE.md
What it produces
- Run the charter conversation
- Scaffold the project folder structure
- Ingest a source document
- Add a KII/FGD to the transcript pipeline
- Write a per-doc summary
- Refresh the three navigation files
Pairs with: research-synthesis-pipeline, qualitative-methods, desk-review-generator, stakeholder-mapping, knowledge-management.
From download to answer in three steps.
Download & unzip
You get a folder with SKILL.md, its references and a cheatsheet.
Install it
Drop it in your skills directory (Claude Code / opencode), or upload SKILL.md to a Project, Custom GPT or Gem.
Ask
Ask in this domain. It answers from the source, cites the section, or says [NOT IN SOURCE].
Before you install.
What exactly is a skill?
A skill is a folder your AI loads on demand: a SKILL.md with the method, reference files, a cheatsheet and a provenance manifest. The model answers from it and replies [NOT IN SOURCE] instead of inventing.
Which AIs does it work with?
This skill installs into Claude Code, opencode, Claude.ai, ChatGPT, Gemini — and any tool that reads a SKILL.md.
Will it answer from the source?
Yes. Every package carries explicit answering rules: cite the section for each claim, label authority vs reasoning, and never invent a figure, citation or rule.
Is it kept current?
Yes. Method and donor skills are reviewed quarterly and you can re-download the latest version from your dashboard.
Research & Evaluation Corpus Setup
Scaffold a navigable, version-controlled corpus for any qualitative research or evaluation project.