About geolocation-skill
geolocation-skill provides a step-by-step method for inferring places from images, for tasks that require checking geographic clues. It combines hypotheses at multiple scales, visual inspection, and web searches, with a capped iteration loop and privacy guidance.
Key features
- 01Extract text, infrastructure, and nature clues
- 02Form hypotheses at multiple scales
- 03Crop and enhance image details
- 04Cross-check findings with web searches
- 05Limit analysis to five iterations
Use cases
- 01Infer a likely country or city from a photo
- 02Verify a location using signs and landmarks
- 03Record location reasoning and confidence
Install
Copy the command and run it in a terminal.
npx skills add https:/After installing, type/geolocation-skillin Claude Code to confirm it loaded.
Bundled files
References and scripts installed along with SKILL.md.
reference/regional-signatures.mdverification-strategies.mdvisual-clues.mdDetails
From the project READMEGeolocation Reasoning
Method: Iterative reasoning loop (max 5 iterations) Inspiration: GeoVista research framework
Core Workflow
1. Visual Clue Extraction
| Category | Look for |
|---|---|
| Text/Signage | Languages, scripts, street signs, license plates, phone formats |
| Infrastructure | Road markings, traffic lights, poles, utilities, bollards |
| Architecture | Materials, styles, roof types, density patterns |
| Nature | Vegetation, terrain, geology, sun position, weather |
| Culture | Vehicles, driving side, clothing, shops, brands |
2. Hypothesis Formation
Form hypotheses at multiple scales:
| Scale | Example |
|---|---|
| Continent/Region | Europe, Southeast Asia |
| Country | Specific nation |
| Region/State | Sub-national |
| City/Town | Locality |
| Precise | Street-level |
For each: Note supporting evidence, confidence (H/M/L), conflicts.
3. Image Examination
When details unclear:
- Crop regions of interest
- Enhance text/signage
- Isolate architectural details
4. Web Search Verification
| Strategy | Query pattern |
|---|---|
| Sign text | "[exact text]" location |
| Features | [feature] [suspected region] |
| Business | "[name]" address |
| Landmark | [description] landmark [region] |
| Road | [designation] [country] |
Cross-reference multiple clues, look for consistent attribution.
5. Iterative Refinement
After each cycle:
- Update hypothesis confidence
- Identify remaining uncertainties
- Determine if more examination helps
- Continue until confident or limit reached
Output Format
**Location Analysis**
Observed Clues:
- [Key observations with confidence]
Reasoning Process:
- [Hypothesis formation and verification]
**Conclusion**
- Most likely location: [As specific as evidence allows]
- Confidence: High/Medium/Low
- Key evidence: [Primary determinants]
- Uncertainties: [What remains unclear]
Important Considerations
| Aspect | Guideline |
|---|---|
| Confidence | High requires multiple independent confirmations |
| Single clues | Can mislead (similar signage in multiple countries) |
| Privacy | Don't identify individuals or residential addresses |
| Iteration | Max 5 cycles, stop when marginal value low |
References
references/visual-clues.md— Clue categoriesreferences/regional-signatures.md— Region-specific indicatorsreferences/verification-strategies.md— Search patterns
🎯 COMPLETED: [SKILL:geolocation-skill] [location identified from image] 🗣️ CUSTOM COMPLETED: [SKILL:geolocation-skill] [Location found]