---
title: "math-proof: solo: review, setup & limitations · Undominated.ai"
canonical: https://undominated.ai/skills/anthropic-math-proof-solo/
description: "Works one hard mathematics problem in the current Claude Code session, writing each settled step to a notes file as it goes and ending with a self-contained proof.md."
---

# math-proof: solo: review, setup & limitations · Undominated.ai

> Works one hard mathematics problem in the current Claude Code session, writing each settled step to a notes file as it goes and ending with a self-contained proof.md.

[← Explore all skills](/skills/)

RESEARCH AND DOCUMENTATION / Anthropic

# math-proof: solo

Works one hard mathematics problem in the current Claude Code session, writing each settled step to a notes file as it goes and ending with a self-contained proof.md.

 [See setup guidance ↓](#setup)[Original source ↗](https://github.com/anthropics/claude-plugins-official/tree/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/skills/solo)

SOURCE REVIEW

 Reviewed 2026-10-21
 Evidence 3 linked sources
 Publisher Anthropic
 Licence [Apache-2.0 ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/LICENSE)
 Revision 41f32e407b45
 [Read what was—and wasn’t—checked ↓](#review)

“Work on one hard mathematics problem in this session yourself, with no sub-agents: reason in stages, record each settled step in a notes file so that nothing written is lost if a response is cut off, and end with a self-contained proof.md. Usage: /math-proof:solo <the problem, stated in full, or the path of a file holding it>.”

 [Anthropic · upstream description ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/skills/solo/SKILL.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - Its tool boundary is written down: the frontmatter disallows WebSearch, WebFetch and AskUserQuestion and pre-approves Read, Write, Edit, Glob, Grep and Bash only for mkdir, cp and cmp, and the body says to use the shell for nothing else.
- It is built around losing nothing to a cut-off response: the plan and every settled step go to DIR/notes.md before further reasoning, a rerun in the same directory continues from those notes, and the end product is a proof.md a referee can read without any other file.

### A good fit for

 - Work a hard mathematics problem in one Claude Code session while keeping a written trail in notes.md
- Resume an interrupted proof attempt by giving the same /math-proof:solo line again in the same directory
- Produce a self-contained proof.md, followed by a reply saying whether it resolves the problem completely or, in proof.md's words, what it leaves open

### Weigh up before choosing

 - Last commit touching plugins/math-proof/skills/solo at revision 41f32e40: 2026-09-30 (aa5654b7).
- Manual invocation only: disable-model-invocation is true, so the skill runs from /math-proof:solo, not from a matching request.
- The README says proof.md is the model's own account and any result it cites from the literature is quoted from memory, so it should be read critically.
- One problem per run directory: if DIR already holds work on a different problem, the skill stops and asks for DIR=<another directory>.
- In the default permission mode Claude Code asks the user to approve each file write; the README's unattended form uses claude -p with --dangerously-skip-permissions.
- The README's settings block applies to every session that reads the settings file it is added to, and the README says to remove it when you stop using the plugin.

02 / THE REVIEW RECORD

## What we actually inspected.

Source review has boundaries. A clear record is more useful than a “safe” badge.

### Material inspected

 - plugins/math-proof/skills/solo/SKILL.md
- plugins/math-proof/README.md
- plugins/math-proof/LICENSE

### Our findings

 - The run directory defaults to ./math-proof-solo; its files are problem.md, notes.md and proof.md.
- A problem given as a file is copied byte for byte with cp; a problem given as text is written exactly as received.
- The SKILL.md tells the session it has no web access and no code execution and that this is a pure reasoning task.
- The skill tells the session to work unattended and never stop to ask questions.

### Not established by this review

 - No problem was run and no proof.md was produced or checked.
- Not executed: it was not installed or run.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

## Add a skill to your workflow.

[Upstream setup instructions ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/README.md)
 - The math-proof README prints `/plugin install math-proof@claude-plugins-official`. It installs both math-proof skills, solo and siege, and the three siege agents, not solo alone, so install.command is null.
- The README then asks for an "env" block in ~/.claude/settings.json, or in .claude/settings.json of the directory you run from; that block is reproduced as configExample.
- Use: in an empty working directory, run `/math-proof:solo <the complete problem statement, or the path of a file holding it>`; `DIR=path` before the problem picks another run folder.
- This review installed nothing and ran no problem.

### Before you start

 - Claude Code 2.1.280 or later, the version the math-proof README requires for both skills

### Documented configuration example

Use the client and authentication context described in the setup guidance. Configuration formats vary between hosts.

 {
 "env": {
 "CLAUDE_CODE_MAX_OUTPUT_TOKENS": "128000",
 "CLAUDE_CODE_DISABLE_BACKGROUND_TASKS": "1",
 "CLAUDE_ASYNC_AGENT_STALL_TIMEOUT_MS": "14400000"
 }
} Copy configuration ↗

### Compatibility

Claude Code

### Writes problem, notes and proof files in one run folder; no web tools

 - allowed-tools: Read, Write, Edit, Glob, Grep, Bash(mkdir *), Bash(cp *), Bash(cmp *)
- disallowed-tools: WebSearch, WebFetch, AskUserQuestion
- Writes DIR/problem.md, DIR/notes.md and DIR/proof.md under the run directory, ./math-proof-solo by default

### Cost model

Apache-2.0 per plugins/math-proof/LICENSE; one long session of the user's model usage, with the README raising the output limit to 128000 tokens per response

04 / FOLLOW THE EVIDENCE

## The source trail.

Our notes are separate from the original resource. Check upstream before adopting a new version.

 - [math-proof solo SKILL.md ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/skills/solo/SKILL.md) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/skills/solo/SKILL.md Supports: upstreamDescription, summary, whySelected[0], whySelected[1], bestFor[0], bestFor[1], bestFor[2], limitations[1], limitations[3], review.findings[0], review.findings[1], review.findings[2], review.findings[3], access.permissions[0], access.permissions[1], access.permissions[2]
- [math-proof README ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/README.md) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/README.md Supports: install.sourceUrl, install.steps[0], install.steps[1], install.steps[2], install.prerequisites[0], configExample, limitations[2], limitations[4], limitations[5], access.cost, compatibility, summary
- [math-proof LICENSE ↗](https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/LICENSE) Checked 2026-10-21 https://raw.githubusercontent.com/anthropics/claude-plugins-official/41f32e407b450055044f167db7ded5e8ff67d203/plugins/math-proof/LICENSE Supports: license, access.cost

KEEP COMPARING

## Other approaches to consider.

Related by category or shared topics. These are alternatives to inspect, not a measured quality order.

 [### math-proof: siege ↗ Runs an hours-long, multi-round attack on one hard mathematics problem with judge and worker sub-agents and a bookkeeping script, ending in a proof.md whose Status section says what is and is not proved.](/skills/anthropic-math-proof-siege/)[### Anthropic Documentation Coauthoring ↗ Guides a human and agent through audience discovery, section-by-section drafting and a separate-reader review of a substantial document.](/skills/anthropics-doc-coauthoring/)[### Claude Academy guide ↗ Answers a Claude how-to first, then may name at most two Academy items copied from a fetched catalog.](/skills/anthropics-academy-guide/)

 [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/)

## Continue your investigation

 - [Choose a toolkit](/tools/)
- [Inspect execution hosts](/agents/)
- [Check tool access](/mcp-servers/)
- [Check project dependencies](/check-repo/)
