classify

A command that labels images as photos, screenshots, documents, or memes.

In plain words
What is it for?
Use it to organize selected images or an entire collection, with filters for folders, dates, locations, file types, and labeled people.
Why use it?
A camera roll often contains more than camera photos, making it difficult to separate useful pictures from receipts, screenshots, and other images.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/erhangundogan/videre/classify
Clone the repo
git clone --depth 1 https://github.com/erhangundogan/videre
Per session 13 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,598 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00013 $0.01598
Opus 5 $0.00006 $0.00799
Sonnet 5 $0.00003 $0.00320
Haiku 4.5 $0.00001 $0.00160

Measured 2d ago against content hash 567ee4061429, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

classify scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

docs/src/content/docs/commands/classify.md · 157 lines

How it starts

The opening of the file, as written. The whole thing — 157 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Tags each image as photo, screenshot, document, or meme, so you can separate real photographs from the receipts, memes and screenshots that accumulate in a camera roll.

videre classify                        # classify everything not done yet
videre classify --reprocess            # redo everything, including already-tagged
videre classify --margin 0.05          # how confident it must be (default 0.05)
videre classify --silent               # no per-image progress
videre classify --db ~/photos.db       # use a specific database
videre classify --model <id>           # classify a specific model's data
videre classify --type image           # only images
videre classify --person Ada           # only photos of a labelled person

Classify takes every filter, so a run can be narrowed by anything already known about a file:

videre classify --type image --after 2025-01-01     # this year's photos only
videre classify --path ~/Photos/Inbox               # one folder
videre classify --location "Berlin, Germany"        # photos taken near a place
videre classify --person "Alice" --reprocess        # re-label one person's photos
videre classify --ext heic --type image --date 2024 # composed: format, kind, year

:::tip These filters work the same way across commands, and combine. See scoping a run. :::

The workflow

It reuses the vectors videre embed already computed, so there is no new model to download and no image is read from disk again. On a library that is already embedded it finishes in seconds to minutes rather than hours.

videre embed                           # prerequisite, the slow part
videre classify                        # fast, reuses that work
videre search --category screenshot

Then use it to clear out the clutter:

videre search --category screenshot -k 5000 > /tmp/shots.txt
videre search --category meme | xargs -I{} mv {} ~/memes/
videre search --category document        # receipts, tickets, forms

Resumable: rerunning only classifies what is not yet done, so adding photos means scan, embed, classify again and only the new ones are considered.

Read the full file on GitHub · 157 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 157 lines · 13 tokens per session scan A 567ee4061429

Subscribe to this mod's changes

classify is a command published in the GitHub repository erhangundogan/videre (2 stars, last pushed 3d ago), licensed Apache-2.0. It adds 13 tokens to every session and 1,598 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.