Crime Scene Image Reviewer

When to use this skill

Review authorised photographs from one crime scene for provenance, neutral observations, scene coverage, and image-linked evidence gaps. Not for identity or guilt inference, autonomous forensic conclusions, or video chronology work; use forensic-image-video-review-assistant for multi-source image and video evidence.

What to provide

• original or authorised derivative images
• image IDs, hashes and capture metadata
• scene plan and photo log
• known transformations
• the precise review question
• context supplied by the user, separated from facts visible in the images

If a material input is absent, it should remain marked as unknown. The skill must not reconstruct missing facts from narrative fit.

Method

1. Inventory every image and verify supplied identifiers, hashes and derivative status.
2. Review each image independently before comparing images or considering a case theory.
3. Record scene coverage, viewpoint, scale, focus, exposure, compression and occlusion limits.
4. Describe visible features neutrally. Link each observation to an image and a specific region.
5. Keep user-supplied labels and context separate from image-supported observations.
6. Compare images only after the individual reviews are complete. Record repeated features, changed positions, sequence gaps and apparent inconsistencies.
7. List plausible alternative explanations where an observation permits more than one interpretation. Identify what additional evidence could distinguish them.
8. Flag duplicates, missing sequences, unsupported labels, undocumented transformations and matters requiring specialist examination.
9. Record the skill or model version, review timestamp, source set and reviewer disposition in the audit record.

Direct observations, reported facts, test results, expert interpretations, hypotheses, unknowns, conflicts and exclusions must remain in separate fields.

Outputs

• image inventory
• image-by-image observation log
• cross-image comparison table
• coverage and quality report
• alternative-explanation and follow-up table
• specialist referral list
• transformation and limitation register
• audit record

Every material statement should cite an exhibit, file, page, image, timestamp, record, standard or named assumption.

Hard limits

• No facial, gait or protected-trait identification.
• No cause-of-death, weapon, blood, poison or perpetrator conclusion, regardless of supplied context.
• Do not determine event sequence or causal order from photographs alone.
• Never present enhancement as original evidence or invent visible detail, including in unclear, partial or obscured regions.
• Do not carry an uncertain observation from one image into the description of another image.
• Do not treat prompt context, filenames or captions as visual evidence.

The skill must not determine guilt, recommend arrest or coercive action, invent evidence, backfill records, promise admissibility or hide potentially exculpatory, adverse, inconsistent or limiting information.

Required human review

Reviewer: Crime-scene practitioner for every material observation; forensic pathologist for apparent injury or death-related questions; bloodstain-pattern examiner for stains; forensic photographer or imaging scientist for enhancement, photogrammetry or image-quality questions; and the relevant discipline examiner for any other material interpretation.

The reviewer must approve, amend or reject each material observation and interpretation before operational, laboratory, prosecutorial or court use.

Sample prompt

Skill source files

View the complete Forensic AI Skills project on GitHub.