Cognitive Technique Mapper

Cognitive Technique Mapper (CTM) pairs real-world problem descriptions with the most suitable cognitive or problem-solving techniques using a configurable CLI and optional FastAPI surface.

Installation

# Python 3.12+ recommended
python -m venv .venv
source .venv/bin/activate
uv pip sync requirements.lock

# Configure provider credentials (example)
cp .env.example .env  # or export directly
export AZURE_API_BASE="https://<your-endpoint>.openai.azure.com/"
export AZURE_OPENAI_KEY="..."
export OPENAI_API_KEY="..."
export ANTHROPIC_API_KEY="..."

Quick Start

# Seed or refresh local data
python -m src.cli refresh --skip-embeddings

# Capture a problem description, analyze it, and generate justification
python -m src.cli describe "I'm torn between two job offers."
python -m src.cli analyze --show-candidates
python -m src.cli explain

# Optional: run the FastAPI service
uvicorn src.api:app --reload

Features

• Configurable workflows (describe, analyze, explain, simulate, compare, feedback, interactive-flow) that blend SQLite data, vector search, and LLM reasoning through litellm.
• Shareable Markdown reports plus preference-aware recommendations backed by feedback loops and history inspection commands.
• YAML-driven configuration (config/settings.yaml, config/models.yaml, config/providers.yaml) for providers, models, and storage paths.
• Technique catalog lifecycle commands (techniques list|add|update|import|export|refresh) that keep SQLite and optional Chroma embeddings synchronized.
• Lightweight FastAPI server mirroring the CLI, enabling programmatic workflow invocation, health checks, and optional GraphQL access.

Developer

For detailed setup, architectural notes, environment variables, and contribution workflows, see README-DEV.md. Release history follows Keep a Changelog; consult CHANGELOG.md for the latest updates.

License

Distributed under the MIT License. See LICENSE for full terms.