context-mode

Compress bulky freelance HTML (Kwork / Weblancer / FL) into agent-ready fragments before it hits the LLM context window.

• Drops nav/footer/similar-jobs noise
• Keeps budget / deadline / description / skills windows
• Zero required deps (optional beautifulsoup4)

Install

pip install -e .
# or after publish:
# pip install context-mode

CLI

context-mode page.html --site kwork
context-mode --url https://kwork.ru/projects/123 --site auto
context-mode page.html --max-chars 3000 --site weblancer

Library

from context_mode import compress, html_to_text

text = html_to_text(open("page.html").read())
print(compress(text, site="kwork", max_chars=4000))

Why

Freelance pages dump 50KB+ of chrome. Agents need title, budget, body, skills, deadline, client stats. This cuts noise before the model sees it.

Layout

src/context_mode/
  __init__.py
  compress.py
  cli.py
tests/
compress_extract.py   # legacy single-file CLI (compat)

Dev

python3 -m pip install -e .
python3 -m pytest tests/ -q

License

MIT · AMEOBIUS-team

Freelance portfolio: https://ameobius-space.github.io/kwork-portfolio/