where-was-this-taken
End-to-end workflow to establish where and when a photo or video was captured and whether it is authentic — evidentiary handling, metadata extraction, reverse image search for provenance, visual geolocation, chronolocation from shadows, and manipulation checks, ending in a location finding with a stated confidence radius. Use when asked to verify where an image was taken, confirm or refute a claimed location or date, or authenticate media before relying on it. Applies to insurance claims, litiga
Security Assessment
Detected risks:
About where-was-this-taken
Where Was This Taken is an end-to-end OSINT workflow for establishing where and when a photo or video was captured and whether it is authentic, producing a location finding with a stated confidence radius, a time window, an authenticity assessment, and an evidence chain for each. It solves the problem of verifying or refuting a claimed location or date before relying on media, and of catching manipulated content so an analyst does not confidently geolocate a composite.
The workflow orchestrates other technique skills in a deliberate order: secure a pristine, hashed, read-only copy of the original (working from the file, not a screenshot, and using yt-dlp for platform video); extract metadata before any transformation because editing rewrites tags, prioritizing GPS fields, camera bearing, and UTC timestamps; establish provenance via reverse image search since the earliest caption often names the place; perform visual geolocation from pixels as an independent test even when GPS is present; and derive time via chronolocation from shadow azimuth and length against a sun-position calculator, corroborated with a weather archive. The order (metadata before analysis, provenance before geolocation, authenticity before commitment) is emphasized as the most important thing in the file.
The skill carries `disable-model-invocation: true` and a strong ethics gate: a Step 1 scope definition and a 'publication floor' decided before the answer is known, explicitly acknowledging that geolocating a private individual's home is the mechanic of stalking and that conflict imagery can endanger people in frame. It targets verification analysts, journalists, insurance and litigation investigators, and human-rights and disinformation researchers.
FAQ
What does it produce?
A location with a confidence radius, a time window, an authenticity assessment, and a documented evidence chain for each conclusion, drawn from metadata, provenance, visual geolocation, and chronolocation.
Why is step order so strongly emphasized?
Because analysis destroys metadata, provenance often reveals the location outright before geolocation is needed, and authenticity checks must come first so effort is not spent geolocating a manipulated composite.
How does it protect the evidence?
It hashes the original with sha256 (and a second algorithm), sets it read-only, works only on named derivative copies, and records the full acquisition provenance including URL, fetch timestamp, and source platform.
What ethical safeguards does it apply?
It requires reading the repo ETHICS.md, naming a legitimate objective, and deciding a 'publication floor' precision before the answer is known, explicitly noting that geolocating a private individual's home is the mechanic of stalking regardless of intent.
Is GPS metadata treated as the answer?
No. GPS is treated as a lead in a writable field; the workflow independently verifies it through visual geolocation and camera-bearing checks, since a confirmed visual match plus metadata is far stronger than either alone.
Install where-was-this-taken
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
useosint/osint-skills