v0.8.0 - Apollo waterfall: intent-gated lookalike expansion

[0.8.0] - 2026-08-12

The Apollo starter learns to grow the list itself. waterfall.py expands a source list with lookalike companies through gated searches that drain from deepest intent to shallowest — job postings for the pain, fresh funding, tech-stack twins, then plain firmographics — so every company arrives tagged with why it made the list. Apollo does not expose buying intent through the API; the gate tag is the intent layer you build yourself.

Added

• waterfall.py: gated lookalike expansion to a target list size. Gates: T0 external evidence (--clearbox CSV), T1 hiring-for-the-pain (job titles + posting recency), T2 fresh funding (amount window as a round-stage proxy), T3 tech-stack twins (verified technology UIDs), T4 firmographic fingerprint. Flags: --target, --double (target = seed-list size), --seed, --dry-run (per-gate market sizing, writes nothing). Output is init_db.py-compatible, so it feeds the existing expand → score → reveal → sheet pipeline directly.
• waterfall_config.example.json: copy to the gitignored waterfall_config.json and tune to your ICP. Optional per-gate cap keeps one deep gate from filling the whole list, so the output stays a mix of intent tiers.
• search_companies() in the starter's Apollo client (mixed_companies/search — company search, no credits drawn).
• Bonus columns: per-company intent_strength (populates when the plan carries intent), six-month headcount growth, and printed revenue land in the output when Apollo returns them.

Fixed

• Staffing-agency noise in job-posting gates: staffing/recruiting firms post roles for their clients and would otherwise flood any hiring gate; they are dropped by NAICS prefix (5613, 541612 by default, configurable).