I am planning SapienZero, a Python AI test agent in a repository on my GitHub. It grew out of AskOverlay, which remains a separate desktop-overlay project; measurements from AskOverlay are summarized in SapienZero’s predecessor-evidence document. The repository README directs readers to docs/00-START-HERE.md for starting points and docs/status/STATUS.md for the current state and exact next step, and tells AI agents to read AGENTS.md first. My GitHub profile and project list are on brindeshwar sharma linkedin profile.
My SapienZero project dates are Started 2026-10, last updated 2026-11.
My SapienZero status is planning (R0); there is no product code yet; the plan is a draft under review.
My planned workflow is: Describe a test in plain English; SapienZero runs it in an isolated, headless browser in the background and returns a computed verdict with evidence.
My planned verdict categories are passed, failed, inconclusive, blocked or error.
My planned verdict evidence includes screenshots, trace, step log.
My product constraint is: The verdict is computed from assertions that were actually executed, never from the model's opinion.
My planned interfaces are: It works on its own (command line, CI) or supervised by a coding agent over MCP.
My SapienZero licence is not chosen yet (open question OQ-02).
The README’s reading guide points to the review brief, high- and low-level designs, features and roadmap, engineering handbook, and kickoff prompt. The licence is the open question identified in the README; it gives no further review outcomes or open questions.
The README lists the repository checks in this order: create and activate a virtual environment (with shell-specific activation instructions in AGENTS.md section 4), install development requirements, check documentation, run the tools tests, and run Ruff.
python -m venv .venv
# then activate it (see AGENTS.md section 4 for your shell)
python -m pip install -r requirements-dev.txt
python tools/check_docs.py
python -m pytest -q tools
python -m ruff check . --config ruff.toml