context-onboard 0.1.0

Ingest, process, and package business context for AI assistants

ContextBase Onboarding Service

Ingest, process, and package organizational context for AI assistants.

context-onboard extracts knowledge from your team's existing tools and surfaces it as structured knowledge graphs, ready-to-use system prompts for Claude/GPT/Gemini, and beautiful HTML reports.

Features

• 🔌 Multi-source ingestion — Slack, Email (IMAP), Google Drive, Airtable
• 🧠 Knowledge graph — Deduplicates entities, merges relationships, computes statistics
• 📝 AI prompts — Generates system prompts for Claude, GPT-4, and Gemini
• 🎁 Deliverable packaging — Creates ZIP archives with graph JSON, prompts, and HTML reports
• 🔧 MCP-style config — Tool-use configuration for Model Context Protocol

Quick Start

# Install
pip install context-onboard

# With all extras
pip install "context-onboard[all]"

Usage

1. Ingest — Pull data from your sources

Create a config file (config.yaml):

output_dir: ./data
ingestors:
  slack:
    token: xoxb-your-bot-token
    enabled: true
  email:
    host: imap.gmail.com
    username: [e-mail]
    password: "app-password"
    max_messages: 200
    enabled: true
  drive:
    token:
      token: ya29...
      refresh_token: 1//...
    max_files: 100
    enabled: true
  airtable:
    api_key: patXXXX...
    base_ids:
      - appXXXX1
    enabled: true

Then run:

context-onboard ingest --config config.yaml

2. Build — Generate knowledge graph and prompts from existing data

context-onboard build --data ./data --company "Acme Corp" --output ./output

3. Package — Create a deliverable ZIP

context-onboard package --data ./output --output ./deliverable --company "Acme Corp"

Architecture

context_onboard/
├── __init__.py              # Package init, version 0.1.0
├── models.py                # Core data models (Entity, Relationship, ContextPackage)
├── cli.py                   # CLI entry point (argparse)
├── ingestors/
│   ├── base.py              # BaseIngestor ABC
│   ├── slack.py             # SlackIngestor (slack_sdk)
│   ├── email.py             # EmailIngestor (imaplib — stdlib)
│   ├── drive.py             # GoogleDriveIngestor (google-api-python-client)
│   └── airtable.py          # AirtableIngestor (requests)
├── processors/
│   ├── knowledge_graph.py   # Deduplication, merging, graph stats
│   └── prompt_library.py    # Claude, GPT, Gemini prompt generation + MCP config
└── output/
    └── package.py           # ZIP packaging + HTML report generation

Data Models

Model	Description
EntityType	Enum: PERSON, TEAM, PROJECT, DOCUMENT, TOOL, PROCESS, GLOSSARY_TERM
Entity	A node in the context graph (id, name, type, description, metadata)
Relationship	A typed, weighted edge between two entities (0-1 strength)
ContextPackage	A complete snapshot from one ingest source or the merged graph

Optional Dependencies

Feature	Install command
Slack	pip install "context-onboard[slack]"
Google Drive	pip install "context-onboard[google]"
Airtable	pip install "context-onboard[airtable]" (requests included by default)
YAML config	pip install "context-onboard[yaml]"
All	pip install "context-onboard[all]"

Email ingestor uses stdlib only (imaplib, email) — no extra deps needed.

Development

git clone https://github.com/contextbase/context-onboard.git
cd context-onboard
pip install -e ".[dev]"
pytest

License

MIT