I build tools
people install.

See what Codex actually burned.

A local CLI that parses Codex sessions and shows token burn by model and by file. 1,474 downloads in 12 months, no telemetry, no auth.

• Parses Codex JSONL on disk , no API call needed
• Groups by model and by file so the costly file stands out
• 7-day and 30-day windows with one flag
• Zero config , runs from npx
• No telemetry, no auth, no network
• Built for the question I kept asking: where did tokens go?

Limitation: only as good as Codex's own session logs. Breaks if the log format shifts.

// read local Codex sessions, sum tokens by model
for (const s of sessions) {
  tokens[s.model] = (tokens[s.model]||0) + s.usage.total;
} // no network, no auth , just files on disk

Live SaaS - open it, run it, keep the data.

Self-serve B2B lead generation you can clone and run today - not screenshots. FastAPI + Next.js 16 with worker, Postgres-backed, batch + cron, ships with make dev / docker-compose up. Verified via /tmp/LeadPilot - full API, auth, and exports, not a mock flow.

Stack: Next.js 16 · FastAPI · Gemini · Apify · Docker · Railway

• Multi-source discovery: Google Maps via Apify + Instagram + email crawler on real domains
• Gemini intelligence: smart scoring + review-grounded outreach - configurable model via env
• Batch + daily cron: python main.py --city "London" --agent --find-emails and run_daily.sh
• Full product: api/ + frontend/ + worker.py - make dev starts all three
• Auth + rate limits + exports: Google login, paginated leads, CSV export, status pipeline
• Live architecture: vercel.json (sin1), railway.toml, docker-compose - one command to run

Limitation: Apify + Gemini are required for live scrape - without keys, guest preview runs in demo mode (GUEST_PREVIEW_MODE=demo).

# clone + run - frontend, api, worker (one command)
git clone https://github.com/Rishet11/LeadPilot.git
make dev  # api :8000 + worker + frontend :3000
# or
docker-compose up --build  # same stack, containerized