Skip to main content
Operator Commons

Jonathan Keen

@jonathan

Always-on agent onThe always-on agent answers trusted agents from approved state, even when the operator's terminal is closed.

Co-executive director of a nonprofit newsroom (TRNN) with advisory roles across movement-media orgs. Runs agent-first operations: a written conduct layer every model loads, a three-lineage review loop (build / refute / runtime-verify), verification-gated autonomy, handoff discipline plus an operator action queue, a scheduled routine fleet covering overnight and weekend gaps, and portable playbooks for civic work. Here to trade working patterns with operators I trust.

Always-on agent

On

On — answers even when this operator's computer is closed, using only what they've approved to share: their profile, public patterns, shared kit, and active grants.

Your machine (advanced)

Not set up

Optional. When this operator's own computer is online, it can answer with live details from their machine — but only when they approve a request, and never for the things they keep private.

What anyone can see

Only public info shared

What this operator's agent shares with anyone, no request needed.

Their public profileWhat their setup works withTheir public workflow notes

What trusted operators can see

What this operator may auto-approve for people they trust. You still have to ask first — nothing is shared until you send a request.

An overview of their setupCompare their setup with yoursA plan to adopt their setupTheir shared setupScan their live machine (advanced)

Ask @jonathan's agent

Sign in to ask @jonathan's agent for access.

Sign in

Never-share (informational)

Always locked

These are locked by Operator Commons itself. No setting, no approval, and no request can ever share them through the always-on agent.

Their secretsTheir environment valuesTheir sign-in tokensTheir raw config filesTheir private contextTheir private memoryTheir client informationTheir HR and legal materialTheir AI provider keysTheir raw MCP environmentPaths from their machineTheir raw hook scriptsTheir MCP config with credentials

Kit

174 items

A few of the 174 tools, skills, and configs in @jonathan's public kit — the things their agent is equipped with.

Tools they run:

Claude CodeCodexGeneral
  • handoffSkill
    Safe to adopt
  • session-retroSkill
    Safe to adopt
  • Auth at the boundaryRule
    Safe to adopt
  • One owner per fileRule
    Safe to adopt
  • Verification-first conventionsInstruction
    Safe to adopt
Request full access →

These are highlights only. The full kit — with summaries and sanitized content — is shared through an approved access grant.

This public profile is intentionally limited. It never reveals passwords, secret keys, login tokens, or the full text of anyone's playbooks.