B2B Prospecting Agent (starter)

Pipeline today: Search → Scrape → Clean → Chunk → Draft (Gemini) → Save to SQLite.

Setup (Python)

Create a virtualenv, then install dependencies:

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

Create your env file:

copy .env.example .env

Set GEMINI_API_KEY in .env.

Run

Full pipeline (requires GEMINI_API_KEY):

python main.py "Acme Corp"

Dynamic prospecting context (optional; prompts adapt to these strings; no hardcoded ICP in code):

• Defaults if omitted: industry B2B Company, job title Decision Maker.

python main.py "Acme Corp" --industry "Healthcare SaaS" --job-title "Chief Medical Officer"
python main.py "Acme Corp" --sequence --industry "Manufacturing" --job-title "Head of Operations"

Skip drafting (only search/scrape/clean/chunk):

python main.py "Acme Corp" --no-draft

Limit URLs:

python main.py "Acme Corp" --top 3

Review workflow:

python main.py --list
python main.py --approve 1
python main.py --mark-sent 1

Data

• Drafts DB is stored at data/drafts.db (created automatically).
• Do not commit .env or data/ (see .gitignore).