本仓库用于系统化沉淀 2026 年高毛利、强防御壁垒的 AI 蓝海商业方向，并通过跨模型协同（Antigravity + Remote Claude Opus 5.5）进行定量可行性评估、立项筛选与 MVP 架构设计。
Three Pay-Per-Event Apify actors backed by one shared, dependency-free core
(permit_intel_core/), built from docs/CLAUDE_ROUTE_A_EXECUTION_PLAN_EN.md.
| Path | What it is |
|---|---|
permit_intel_core/ |
Shared engine (Python standard library only; the apify SDK is optional). It contains the source registry (sources/registry.yaml), the Socrata client, normalizers (E.164 phones, address, v1 record), the spec extractor, license joins, delta state, and JSON Schemas. |
actors/permits-contractor-leads/ |
Actor #1, Building Permits & Contractor Leads. PPE events permit-record and contact-enriched-record. |
actors/contractor-license-radar/ |
Actor #2, Contractor License, Insurance & Bond Expiration Leads (WA L&I). PPE events contractor-record and expiring-alert-record. |
actors/permit-history-lookup/ |
Actor #3, Permit History API. Runs in batch mode or as a Standby HTTP /lookup + MCP /mcp server. PPE event lookup, charged on matches only. |
scripts/probe_sources.py |
Live endpoint, field and freshness probe. Writes probe_report.md. |
scripts/coverage_report.py |
Live fill-rate measurement and the Day-2 gate. Writes COVERAGE.md and refreshes the actor READMEs. |
scripts/sync_core.sh |
Vendors the core into each actor before apify push. --check verifies the copies are current. |
scripts/record_fixtures.py |
Re-records the privacy-scrubbed offline fixtures. |
assets/store_metadata.json |
Store title/description/seoTitle/seoDescription for all 6 Actors, with the search-ranking rationale. tests/test_store_assets.py keeps it in sync with every actor.json. |
scripts/generate_actor_icons.py |
Deterministic Pillow generator for the 6 Store icons in assets/icons/ (512x512 RGBA, 1024 supersampled). --preview out.png renders a dark/light contact sheet; --check only verifies. |
PRICING.md |
Competitor benchmark (live Store API) and the exact PPE configuration. |
templates/ |
n8n workflow, Make blueprint, Clay + MCP setup, and 5 community posts. |
tests/ |
Golden-set spec precision, normalization/schema/client/delta tests, and local runs of all 3 actors. |
python3 -m unittest discover -s tests -v # offline: fixtures only
python3 scripts/probe_sources.py # live: writes probe_report.md
python3 scripts/coverage_report.py # live: writes COVERAGE.md
./scripts/sync_core.sh && cd actors/permits-contractor-leads && apify pushOptional: export SOCRATA_APP_TOKEN=.... The client then sends it as X-App-Token for higher rate limits. On Apify, set it as a secret environment variable in the Console.
Three self-contained Pay-Per-Event actors, built from docs/CLAUDE_APIFY_TOP5_EVAL_REPORT_EN.md. Unlike the permit actors, they have no vendoring step: each actors/<name>/ directory is its own Docker build context.
| Path | Apify name | PPE events |
|---|---|---|
actors/ats-jobs-hiring-signals/ |
ats-jobs-hiring-signals-api: jobs and company hiring signals from Greenhouse, Ashby, Lever, SmartRecruiters, Recruitee and Workday public board APIs |
job-record, company-hiring-summary |
actors/google-ads-transparency/ |
google-ads-transparency-advertiser-leads: advertiser checks and ad creatives from the Google Ads Transparency Center RPCs, with a labeled, never-charged verified-snapshot fallback |
advertiser-check, ad-creative |
actors/keyword-volume-seo-intel/ |
keyword-search-volume-cpc-difficulty-api: modeled volume, CPC, difficulty and intent from live Google/Bing/YouTube autocomplete, with optional bring-your-own DataForSEO for exact data |
keyword-metric |
src/runtime.py is shared: each actor ships an identical copy, and a test fails if the copies drift.
pip install -r actors/ats-jobs-hiring-signals/requirements.txt pytest jsonschema
python -m pytest tests # offline: recorded real responses, Apify meta-schema checks, real-SDK PPE runs
LIVE_TESTS=1 python -m pytest tests # plus live integration tests against the real upstreams
cd actors/ats-jobs-hiring-signals && python -m src.main --local-input '{"companies": ["greenhouse:stripe"]}'
python scripts/record_ads_snapshot.py # re-record the Ads Transparency verified snapshot (live, paced)
python scripts/record_new_actor_fixtures.py # re-record the offline ATS/keyword fixtures (live)docs/BLUE_OCEAN_PROPOSALS_ZH.md：7 大 AI 蓝海项目合集结构化总结（中文完整版，含痛点、产品形态、定价模型、壁垒与 2 周 MVP 切入点）。docs/CLAUDE_FEASIBILITY_PROMPT_EN.md：发往远端 Claude Opus 5.5 的全英文可行性分析任务书与定量评分卡标准（100% English Prompt）。docs/CLAUDE_FEASIBILITY_ANALYSIS_EN.md：远端 Claude Opus 5.5 (--effort high) 输出的 443 行完整英文可行性、单元经济模型与防坑分析报告。docs/CLAUDE_FEASIBILITY_ANALYSIS_ZH.md：Claude 深度可行性报告的中文精译、避坑拆解与 14 天 MVP 落地作战指南。docs/P1_P6_OPERATIONAL_PLAYBOOK_ZH.md：P1 与 P6 深度落地实操手册（做什么、如何做、卖给谁、地域群体、如何对接、公司注册与跨境收款全流程）。docs/SOLO_ZERO_TOUCH_BLUE_OCEANS_ZH.md：单人可做、免商务对接、免注册公司的 5 大 AI 蓝海变现方向（平台托管、MoR 个人收款与投币式自助变现）。docs/CLAUDE_SOLO_ZERO_TOUCH_PROMPT_EN.md：发往远端 Claude Opus 5.5 的 S1–S5（单人·免对接·免开公司）可行性评估英文任务书。docs/CLAUDE_SOLO_ZERO_TOUCH_ANALYSIS_EN.md：Claude Opus 5.5 (--effort high) 对 S1–S5 模式的完整英文评估报告（含 Schema 设计与 14 天排期）。docs/CLAUDE_SOLO_ZERO_TOUCH_ANALYSIS_ZH.md：Claude 对 S1–S5 评估报告的中文精译、避坑拆解与permit-core黄金技术栈指南。docs/CLAUDE_S1_PERMITS_DEEP_DIVE_PROMPT_EN.md：发往远端 Claude Opus 5.5 的 S1（美国施工许可与承包商情报节点）专项研究英文任务书。docs/CLAUDE_S1_PERMITS_DEEP_DIVE_EN.md：Claude Opus 5.5 (--effort high) 输出的 662 行 S1 专项深度研究英文原版报告。docs/CLAUDE_S1_PERMITS_DEEP_DIVE_ZH.md：S1 专项深度研究中文精译 + Apify Store 官方 API 实时查验数据交叉核对报告。docs/CLAUDE_ROUTE_A_EXECUTION_PLAN_EN.md：路线 A 终极落地作战蓝图英文原版（Claude Opus 5.5 + Antigravity 联合制定）。docs/CLAUDE_ROUTE_A_EXECUTION_PLAN_ZH.md：路线 A 终极落地作战蓝图中文完整版（含 3 台售货机矩阵、分工表、90 天里程碑）。scripts/run_kr_claude_feasibility.py：第一轮（P1–P7）远端 Claude 评估启动脚本。scripts/run_kr_claude_solo_eval.py：第二轮（S1–S5 单人免对接模式）远端 Claude 评估启动脚本。scripts/run_kr_claude_s1_deep_dive.py：第三轮（S1 专项深挖）远端 Claude 评估启动脚本。.remote-tasks/：存放远端 Claude 评估输出结果与运行日志。