Lay out a numbered plan before a multi-step task
No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.
What counts as done
An ordered list of thoughts with numbers and a stated total, each call returning promptly, and the host carrying on to its next tool call after the final thought without hanging.
Measured, for our own tests: a 5-thought plan costs under 3,000 tokens of tool traffic beyond the thoughts themselves. No docs we read state these figures, so they don't count against a listing here.
How agents ask for it: think this through step by step first
, plan the change before you edit
, work out the steps before trading
Does the job, with a limit 2
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Clear Thought 1.5 E Waldzell AI | sequential_thinking takes thoughtNumber and totalThoughts only inside a free-form parameters object the schema doesn't describe, nextThoughtNeeded defaults false, and a prompt containing 'options' or 'branch' silently dispatches tree_of_thought. Responses echo the thought plus session context; 100 thoughts per session. source |
| Sequential Thinking (MCP reference server) BB Model Context Protocol, a Series of LF Projects, LLC | Numbered thoughts with totalThoughts, in-memory and returning at once; 2.x coerces string numbers and requires nextThoughtNeeded as declared. Hang and crash reports from Claude Code and Augment (acc-11, acc-12) were closed not planned in October, and 2.x needs Node 24. source |
What stops agents finishing
- host hangs or crashes after a thought
- declared schema differs from runtime validation
- server will not start in the host
Around it
After: code.edit-files, search.web-search.
Where this comes from
Weight 40 of 100 in Reasoning scaffolds: 13 of 28 accounts; the use named in every Hacker News account and in the hang and crash reports from Claude Code, Augment and Cursor
The same as data: /api/v1/use-cases/reasoning.plan-before-acting.json, or the MCP tool get_use_case with id reasoning.plan-before-acting. Other jobs in this category: /api/v1/niches?category=reasoning.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.