Search across the applications you have access to, or ask a question about your work.
What agents may do on their own, on every project. A project can be stricter, never looser.
Agents may create tasks
Agents may move tasks through statuses
Agents may assign people
Agents may close tasks with evidence
Agents may ask decisions
Agents read every new comment before they act
Irreversible steps wait for approval
Agents may message clients
A trigger, an AI step, and a person who checks. The AI step follows the rules above.
Client request becomes a task
Ran 3 times this weekReviewer before approval
Ran 5 times this week · 1 flaggedTell the client when work is done
OffSigned offer becomes a plan
Last ran Mon 10 AugOnce per person. Your Claude then reads and writes Work with your permissions, from any session.
Create your personal token
Add Work to Claude Code
claude mcp add --transport http spark-work https://spark.estorly.com/mcp --header "Authorization: Bearer <your token>"Check the connection