KATA Markdown™ toolkit for human-AI collaboration: schema, validation, and template in a single file
Project description
gospelo-kata — KATA Markdown™ for Human-AI Collaboration
A document format and toolkit designed for human-AI collaboration. KATA Markdown™ embeds schema, data, and template in a single file — readable and actionable by both humans and AI without special instructions.
Why gospelo-kata?
KATA Markdown™ is designed so that both humans and AI can read, understand, and work with the same document. The format is self-describing: AI can understand the template structure from the embedded schema and prompt without needing external instructions, while humans can read and edit the same file naturally. It acts as a harness for autonomous AI — schema, validation, and trust management guide AI output along safe, structured paths.
When generating documents with AI, you often face these problems:
- No structure — AI outputs free-form text that's hard to validate or reuse
- No round-trip — once rendered, you can't extract the original data back
- No validation — schema violations go unnoticed until review
- AI needs coaching — you have to explain the output format every time
gospelo-kata solves this with a single .kata.md file that contains everything: schema definition, structured data, and a Jinja2-compatible template (built-in engine, no external dependency). The embedded **Schema** and **Prompt** blocks let AI understand the template on its own — no separate instructions needed. Rendered output preserves data bindings via data-kata annotations, enabling round-trip extraction and automated validation.
Features
- Human-AI collaborative format — both humans and AI can read, edit, and generate from the same file
- Self-describing templates — embedded
**Schema**and**Prompt**blocks let AI understand the template without external instructions - Single-file format — schema, data, and template in one
.kata.mdfile - YAML shorthand schemas — define types concisely (
string!,enum(a,b,c),items[]!:) - Round-trip — extract structured data back from rendered documents
- Lint — validate both templates and rendered output (20+ rules)
- AI-friendly —
assemblecommand lets AI generate only YAML data; the toolkit handles the rest - Secure packaging — KATA ARchive™ (
.katar) bundles templates with integrity verification and trust management - Multi-format output — Markdown, Excel, and HTML
- VSCode extension — real-time lint, hover info, preview CSS
- No external templating dependencies for core features (built-in Jinja2 3.1.6-compatible engine; PyYAML required, openpyxl optional for Excel)
KATA ARchive™ (.katar) — Secure Template Packages
KATA ARchive™ is a secure, single-file template package format. A .katar file bundles template, schema, prompt, manifest, and optional images into a ZIP archive that AI can autonomously discover, understand, and use — no manual instructions needed.
- Self-contained — everything AI needs is in one file
- Integrity-verified —
packcomputes a hash stored inmanifest.json; tampered packages are rejected on load - Sandboxed — only
manifest.json,_tpl.kata.md, and image files are allowed; all other file types are blocked - Trust-managed — AI prompt execution requires explicit user approval; changes to prompts trigger re-confirmation
- Structure integrity — rendered
.kata.mdoutput embeds a hash to detect post-render tampering of template structure
# Create, pack, and use a template
gospelo-kata pack-init ./my_template/
gospelo-kata pack ./my_template/ -o my_template.katar
gospelo-kata init --from-package my_template.katar
See the Template Package Guide for details.
Installation
pip install gospelo-kata
# With Excel support
pip install gospelo-kata[excel]
Requires Python 3.11+.
Quick Start
1. Create a document from scratch
cat > todo_tpl.kata.md << 'EOF'
**Prompt**
```yaml
This template generates a task checklist.
Describe items in the items array with task and done fields.
```
# {{ title }}
| Task | Done |
|------|:----:|
{% for item in items %}| {{ item.task }} | {{ item.done }} |
{% endfor %}
<details>
<summary>Schema Reference</summary>
**Schema**
```yaml
title: string!
items[]!:
task: string!
done: boolean
```
**Data**
```yaml
title: Sprint Tasks
items:
- task: Set up CI pipeline
done: true
- task: Write API tests
done: false
- task: Deploy to staging
done: false
```
</details>
EOF
gospelo-kata render todo_tpl.kata.md -o outputs/todo.kata.md
gospelo-kata lint outputs/todo.kata.md
2. Use a built-in template
# List available templates
gospelo-kata templates
# Initialize a project
gospelo-kata init --type checklist -o ./my-project/
3. AI workflow (recommended)
AI generates only the YAML data; build combines it with a built-in template:
# 1. Check prompt + schema
gospelo-kata export checklist --part prompt,schema
# 2. AI creates data.yml following the schema
# 3. Validate, build, and lint
gospelo-kata import-data checklist data.yml -q
gospelo-kata build checklist data.yml -o outputs/
gospelo-kata lint outputs/checklist.kata.md
4. Extract data (round-trip)
gospelo-kata extract outputs/checklist.kata.md -o extracted.json
Reconstructs the original structured data from the rendered document.
KATA Markdown™ Format
A _tpl.kata.md file has four blocks:
**Prompt**
```yaml
Generate a security checklist with categories and items.
Each item needs an id, status (draft/pending/approve/reject), and tags.
```
# {{ title }}
{% for cat in categories %}
## {{ cat.name }}
| ID | Status | Tags |
|----|--------|------|
{% for item in cat.items %}| {{ item.id }} | {{ item.status }} | {{ item.tags | join(", ") }} |
{% endfor %}
{% endfor %}
<details>
<summary>Schema Reference</summary>
**Schema**
```yaml
title: string!
version: string
categories[]!:
id: string!
name: string!
items[]!:
id: string!
status: enum(draft, pending, approve, reject)
tags: string[]
```
**Data**
```yaml
title: Security Checklist
version: "1.0"
categories:
- id: auth
name: Authentication
items:
- id: auth-01
status: draft
tags: [web, api]
```
</details>
Schema Shorthand
| Notation | Meaning |
|---|---|
string |
Optional string |
string! |
Required string |
integer, number, boolean |
Typed values |
int, float, bool, str |
Aliases (→ integer, number, boolean, string) |
enum(a, b, c) |
Enumeration |
string[] |
String array |
items[]!: |
Required array of objects (indent children) |
Rendered Output
gospelo-kata render produces annotated Markdown:
<span data-kata="p-{path}">value</span>— data bindings<div data-kata-each="collection">— loop markers<details>section with Schema + Data for reconstruction
Built-in Templates
| Type | Description |
|---|---|
checklist |
Structured checklist with categories, status tracking, and automation levels |
test_spec |
Test case specification with prerequisites and expected results |
agenda |
Meeting agenda with decisions, action items, and time allocation |
CLI Commands
| Command | Description |
|---|---|
templates |
List available templates |
init |
Initialize project from a template |
render |
Render a .kata.md template to annotated output |
assemble |
Combine built-in template + data file into _tpl.kata.md |
lint |
Validate templates and rendered documents |
extract |
Extract structured data from rendered output |
validate |
Validate JSON/YAML data against a schema |
generate |
Generate Markdown/Excel/HTML from JSON data |
pack |
Pack a template directory into a .katar archive |
pack-init |
Scaffold a new template directory |
export |
Export template parts (prompt, schema, data, body) |
import-data |
Validate data.yml against template schema |
fmt |
Auto-format data-kata spans |
coverage |
Analyze checklist coverage |
edit |
Browser-based data editor |
workflow-status |
Track pipeline progress |
See the CLI Reference for full details.
AI Integration
gospelo-kata is designed to work with AI assistants. The build command minimizes what AI needs to generate — just YAML data following the schema.
Supported AI tools:
- Claude Code — skill files in
skill/claude-code/ - GitHub Copilot Chat — instructions via
.github/copilot-instructions.md
The 3-step workflow (data.yml → assemble → render + lint) works reliably even with smaller models that have limited context windows.
VSCode Extension
Install from VS Marketplace. The kata-lint extension provides:
- Real-time lint diagnostics in the Problems panel
- Hover information for
data-kataattributes - Preview CSS for kata-specific styles
See the VSCode Integration Guide.
Documentation
- Quick Start
- CLI Reference
- KATA Markdown™ Format
- Lint Rules
- Skill Guide
- VSCode Integration
- Copilot Setup
- Template Package
- Prompt Design Guide
- Type Conversion
License
MIT — free for commercial use. Documents generated by this software and templates created by users are the intellectual property of their respective creators. When used with AI services, data may be transmitted to AI providers. See LICENSE.md for details.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file gospelo_kata-0.4.0.tar.gz.
File metadata
- Download URL: gospelo_kata-0.4.0.tar.gz
- Upload date:
- Size: 103.7 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.8
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
09f944cda40014be4b6955feb856b8a1380623dbf343222d93c859ae9a9a3c1a
|
|
| MD5 |
09479b050c2ce7e4d2123e01124e38ad
|
|
| BLAKE2b-256 |
bb886898ab30cf5a2e60ff570fc937218758c44df3f06c656d1ec4fabc911a0d
|
File details
Details for the file gospelo_kata-0.4.0-py3-none-any.whl.
File metadata
- Download URL: gospelo_kata-0.4.0-py3-none-any.whl
- Upload date:
- Size: 112.3 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.8
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
281efc560a9023bd4bcc95099baa33c5543153684b7a4c231af8053a3798ed6e
|
|
| MD5 |
7e634a3773d42087bcfb21ce1cde7176
|
|
| BLAKE2b-256 |
ecc001b86e2e7479467df2eb095ffa829e4adac25afa199285e9ec7188a8ef60
|