Join the founding panel, install local hooks, and automatically contribute privacy-safe receipts.
Quickstart
Get contributing in four steps. You need Node.js 20+ and a founding-panel invite.
Install the CLI
Install the beta from its public GitHub release:
npm install --global https://github.com/bflynn4141/pollen/releases/download/cli-v0.1.0-beta.6/pollen-cli-0.1.0-beta.6.tgzJoin and install hooks
Join before setting up World ID so the server-issued pseudonymous identity is the one you later verify:
pollen join <invite-code>Install hooks for detected Claude Code and Codex installations, then check the result:
pollen setup --agents
pollen doctorWallet and World ID setup are separate and optional for capture. They are required only for payout eligibility:
pollen wallet
pollen verifySetup preserves existing hooks, and detailed observations stay local in
~/.pollen/local.db. Founding-panel participation does not guarantee a payout
while payout automation remains in beta.
Inspect locally
After a completed agent session, inspect exactly what Pollen captured:
pollen myThis reads only your local SQLite database and performs no upload.
Contribute automatically
Completed sessions are placed in a durable local outbox and uploaded in the background. Failed uploads retry at later session boundaries. You can preview eligible receipts or force a retry:
pollen sync --dry-run
pollen syncThe dry run prints aggregate counts and uploads nothing. The client sends only
coarsened fields, including canonical public MCP server/tool identifiers plus
success and latency buckets. Unknown or custom MCP aliases become private.
The server rejects prompts, arguments,
source, files, shell output, transcript paths, credentials, and unknown fields.
Contributor controls
pollen pause
pollen resume
pollen leave --delete-network-dataPause stops local capture without deleting anything. Leave is explicit and permanent: it deletes the contributor's server receipts and revokes the token, then recomputes published aggregates. Local history remains available.
Next steps
- CLI command reference — all available commands
- Trends API — query the aggregated data