Integrations
Works with the AI you already use.
Connect Claude, ChatGPT or your own agent, and ask for any photo in plain words.
Works with
Pick the one you already have open.
Every photo carries the date, the place and the AI's labels for what is in it. That is what these tools read, so a request like “the bartender pouring at night” comes back with the right shot.
Claude
Ask Claude for a photo in plain words. It searches the library, lists what is new and hands you a download link.
How to connectAdd Team Capture as a custom connector. Paste the address from Agent access.
Open Agent access
ChatGPT
The same three tools inside ChatGPT: search by description, list recent photos, get a link to one.
How to connectCreate a connector under Settings, Apps and Connectors. Paste the same address.
Open Agent accessClaude Code
Runs the command line tool from inside a coding session. Find a shot, pull it, do the work, push the result back.
How to connectInstall the CLI on the machine it runs on and log in once with a token.
Set up the CLICodex
Codex can run a shell command, so it can run the CLI. The whole library is four verbs away.
How to connectInstall the CLI on the machine it runs on and log in once with a token.
Set up the CLI
Cursor
Give Cursor's agent your event photos through its terminal. Search, pull and push without leaving the editor.
How to connectInstall the CLI on the machine it runs on and log in once with a token.
Set up the CLIYour own agent
An OpenRouter setup, a pipeline, a script you wrote this afternoon. If it can run a command or make an HTTP request, it can read the library.
How to connectCreate a token under Agent access, then use the CLI or the API.
See both ways in
Or build your own
Two ways in, one token.
Both take the same token from Agent access and both see exactly one account. Revoke the token and both stop.
The command line
One file, no dependencies. Four commands: ls, search, pull and push. Any tool with a shell can drive it.
How to connectnpm install in one line, then log in with a token.
Install itThe API
The same four operations over HTTP. List, search, get a download link, write results back.
How to connectSend the token as a bearer token. Every call is scoped to one account.
See the four calls
The API
Four calls, and that is all of it.
A bearer token gets you the three read calls. Writing results back takes a second scope on the same token. Anything that can make an HTTP request can do all four.
- List the library. Filter by album, by tag and by approval state. Returns metadata and what renditions exist, never the bytes.
- Ask for it in words. Every photo carries the AI's labels for what is in it, alongside the date and the place, so a query comes back in relevance order rather than in upload order.
- Pull one file. A signed, expiring link to the original, the web-sized version, or the corrected one.
- Write your results back. Whatever your agent makes from a photo, a crop, a caption, a cutout, is stored against that photo rather than in a folder somewhere else.
Every call is authenticated, scoped to one account, and recorded in an audit table. No call ever crosses from one account into another.
Rather not write the HTTP yourself? The command line tool is the same four operations, installed in one line.
Access
Point it at your own events.
Tell us what your events look like and what you would want an agent to do with the photos, and we will tell you whether this fits and what it costs.
Start free trial