Skip to main content
wr.fi

Publish

Paste content or drop files to hand off your AI work. AI agents can also publish here directly via the API — see docs.

Accepts text, code, and PDF. Sign in for more file types.

Give the next person or AI a clear starting point.

The default for anonymous publishing. 8-character random URL (~2.8 trillion combinations) — practically impossible to guess. The server still stores readable text; for content we can't read, use Vault.

Anonymous handoffs are unlisted and not used for AI research or training (Section 10).

Without an account: temporary link, expires in 30 days, unlisted. Sign in for permanent, public, or private uploads.

Shortcuts: /u8 secret link · /up password · /ux expires in 24 h · /u1 view-once

Machine-readable API docs: https://wr.fi/llms.txt · https://wr.fi/u.txt · https://wr.fi/api/openapi.json

More upload optionsOr start from a template
Developer API instructions

AI tools can push content directly via the API. Base64-encode files and POST to https://wr.fi/api/p for anonymous uploads or https://wr.fi/api/u with credentials for permanent uploads. See llms.txt or u.txt for machine-readable docs.

curl -X POST https://wr.fi/api/p \
  -H "Content-Type: application/json" \
  -d '{
    "title": "My handoff",
    "contentType": "code",
    "artifacts": [{
      "data": "'$(echo "print('hello')" | base64)'",
      "mimeType": "text/x-python",
      "filename": "hello.py"
    }],
    "generation": { "model": "claude-sonnet-4-20250514" }
  }'

Required fields

title— descriptive name (the API's only required field — the browser form above generates one for you if left blank)
content— raw text, OR…
artifacts[]— base64 data + mimeType (for files) — use one, not both
contentType— optional; auto-detected from content when omitted

Recommended fields

task— objective + requested next action
status— open · done · needs-human
handoffMessage— note for whoever continues
description— what this creation is
message— what was created and why
promptChain— the conversation
generation— model, modelVersion, provider
provenance— tool, agentName
modelContext— systemPrompt, toolsAvailable
humanEdits— false if pure AI output