Intention

A solo-company system for turning intent into action.

Stack

• Language: Python
• Package Manager: pip
• Tools: git
• Execution System: python -m agent_scripts.run (deterministic, intent-based)

Setup

1. Clone the repo.
2. Create a virtual environment: python -m venv .venv
3. Activate: . .venv/bin/activate
4. Install dependencies: pip install -r requirements.txt
5. Set up environment: copy .env.example to .env (if exists) and fill in values.
6. Run: python -m agent_scripts.run --intent "your intent here" --dry-run

Framework Runner

Run latest inbox entry:

python -m agent_scripts.run --once

Run explicit intent with explicit workflow/role:

python -m agent_scripts.run --intent "..." --workflow "Friction → Fix" --role "Architect"

Dry run + print plan (no files written):

python -m agent_scripts.run --intent "..." --print-plan

Usage

• Use intent/docs/intent_inbox.md to capture raw intent (append-only).
• Run the framework runner to produce artifacts in intent/docs/artifacts/.

Project Facts

• Purpose: Turn user intent into executable actions via AI.
• Structure: Minimal, documentation-first, memory-compounding.
• Workflows: See intent/docs/workflows.md for execution paths.

TODO

• Add a license (e.g., MIT).
• Add .env.example for environment variables.
• Add CI/CD pipeline (e.g., GitHub Actions).
• Add CHANGELOG.md for audit findings.