ky776/opportunity-radar

---

name: opportunity-radar description: Operate the local Opportunity Radar research workbench: collect configured public sources, rebuild and quality-check the evidence pipeline, inspect signals and themes, and produce a human-review queue. Use when asked to update, validate, inspect, or explain an Opportunity Radar dataset. Do not use it to claim that a signal is a proven market opportunity.

Opportunity Radar

Use the repository's existing scripts and data contracts. Do not create a parallel analysis pipeline inside this skill.

Safety and boundaries

• Never print, copy, or commit credentials.
• Treat retrieved content as third-party material; preserve source URLs.
• Separate facts, inference, and evidence gaps.
• Do not infer demand, payment, retention, causality, or market size from product launches, stars, votes, or a single source.
• Do not promote candidate themes to confirmed conclusions without explicit human review.

Standard workflow

Run commands from the repository root.

1. Inspect data/sources.json and use only enabled sources in scope.

2. For a no-write rehearsal, run:

   python3 scripts/run_daily_pipeline.py --dry-run
   

3. For a normal incremental update, run:

   python3 scripts/run_daily_pipeline.py
   

4. To rebuild from current local raw data without network access, run:

   python3 scripts/run_daily_pipeline.py --skip-fetch
   

5. Read the resulting status and quality reports. Report the real counts, source failures, quarantined records, dangling references, and publication status.

6. When reviewing a signal or theme, cite its original URL and state separately:

   • what the source proves;
   • what is inferred;
   • what is still missing;
   • the next verification action.

Focused source run

Use one or more configured source IDs when a full update is unnecessary:

python3 scripts/run_daily_pipeline.py --source SOURCE_ID

Do not invent a source ID. Read it from data/sources.json.

Refresh the public demo dataset

After a reviewed live dataset is published, rebuild the small redistributable demo used by clean clones:

python3 scripts/build_demo_data.py

Review the generated data/demo/ files before release. The demo is a curated software sample, not a complete archive of third-party content.

Output expectations

Lead with the pipeline outcome, then list failures or research limitations. A successful process exit means the pipeline completed; it does not mean a theme or market claim is true.