Explain it once, not to every agent.
Claude Code, Cursor, Codex and Grok read the same plan for each job (the spec). Switch agents without starting over.
For engineers
Know when a job is really done.
Each plan ends with a short checklist (Done When). Every item gets checked and leaves a receipt.
What changes
"Done" from an agent used to mean you go and check.
No two agents doing the same job.
One agent holds each job. If it gets stuck, it hands over with notes on what it tried.
Hear what your agents found, without reading every chat.
Agents file what they spot in the same list, rated from broken to minor. Your rating wins.
The agent says it's done. Here's how you know.
You open the job. Each line of the Done When has a receipt next to it: the test that passed, the staging page where you can see the change. One line couldn't be checked, so it stays unchecked and says so. You know what's finished, and what isn't, without taking anyone's word for it.
You make the call.
Annsa ranks and suggests. You decide what ships, with the receipts in front of you. We build Annsa this way: 153 jobs shipped in 12 days with four agents on one list.
01
Evidence lands
Customers · roadmap · ideas · agent findings
02
Annsa ranks it
Who asked · what they said
03
You choose
The reason stays with it
04
You or your agents write the spec
Claude Code · Cursor · Codex · Grok
05
You learn what worked
Checked · decided · ruled out
The product
See the loop inside the product.
Priorities ranked in the middle, the spec they became on the right, every number tracing back to the customers who asked.
Questions people ask.
Q1
Does Annsa change my code?
No. Your agents write the code in your own tools. Annsa holds the spec, the checklist and the receipts.
Q2
What counts as a receipt?
Something you can follow: a test that passed, a staging URL or another link that shows the line is true. A line nobody could check stays unchecked, and Annsa says so.
Q3
Who writes the Done When?
Whoever writes the spec: you, your agent or Ask (Annsa's own agent). It's written before the work starts, so done means the same thing to everyone.
Q4
Can I use it with Claude Code, Cursor, Codex and Grok?
Yes. Each connects over MCP with nothing to install and signs in through your browser the first time. See /docs/connect-claude-code, /docs/connect-cursor, /docs/connect-codex and /docs/connect-grok, then /docs/working-with-specs.
Done means done, with receipts.
Begin with one agent and the work that matters next.