MATHEMATICS / Anthropic
Math-proof worker
Answers one self-contained mathematics question into a file, separates proved claims from sketches and open questions, and appends the task's end marker only when finished.
“A math-proof worker: it answers one self-contained question from a file, writing its answer to a file as it reasons. It has no memory between questions and is launched only by the math-proof plugin's siege skill.”
01 / THE REASONING
Why this made the selection.
- The answer file has to separate what is proved, what is only sketched or conjectured, and what remains open, and a literature result must be labeled as cited from memory.
- The last line is the task's exact end marker, written only when the answer is finished and moved back to the end if the file is extended. The reply abstract is at most 300 words.
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/agents/math-proof-worker.md
- plugins/math-proof/LICENSE
- plugins/math-proof/README.md (installation, prerequisites, use, resume and warnings)
- plugins/math-proof/skills/siege/SKILL.md (frontmatter, setup and orchestration rules; not a full protocol audit)
- Round 2: pinned definition, output contract and permission/dependency boundaries reread on 2026-10-14
Our findings
- The licence file is the Apache License, Version 2.0. The appendix line is Copyright [yyyy] [name of copyright owner].
- Tools are Read, Write, and Edit. model is inherit, effort is low, and maxTurns is 80. The worker reads the query, the problem statement, the problem file if the pasted copy looks garbled, and a named unfinished answer file. Nothing else in the directory is in scope.
- The worker writes intermediate statements before long reasoning so an interrupted attempt leaves recoverable evidence.
- The inspected siege rules launch a fresh worker for a partial answer, retain a partial file, and use ledger.py answers to classify completion before deciding on a rerun.
- The last-line marker is supplied per query; the source requires it even when the worker cannot settle the task.
- A resumed worker must check reusable material itself; an unfinished answer is evidence to inspect, not inherited truth.
- The final answer separates proved statements, sketches and open questions, and literature recalled from memory must be identified as such.
Not established by this review
- The agent was not executed; model behaviour, host compatibility and task outcomes were not runtime-tested.
- No plugin installation, proof run or bookkeeping-script execution was performed.
- The remaining siege protocol and ledger.py implementation were not fully audited.
The review applies to the material and revision named here. A newer upstream release can change its behavior.
03 / PUT IT TO WORK
Use the role in your project.
- Copy the worker file at the reviewed commit into the location your host uses for agent prompts.
- No install command is published. Installing the whole math-proof plugin was not reviewed.
Before you start
- One query file and the problem statement
- A path for the answer file
04 / FOLLOW THE EVIDENCE
The source trail.
Our notes are separate from the original resource.
Check upstream before adopting a new version.
- Math-proof worker ↗Checked
https://github.com/anthropics/claude-plugins-official/blob/b78ac49cdc6b3d7b61c4439470e311f4291265b1/plugins/math-proof/agents/math-proof-worker.md
Supports: summary, whySelected, output format
- Apache-2.0 licence ↗Checked
https://github.com/anthropics/claude-plugins-official/blob/b78ac49cdc6b3d7b61c4439470e311f4291265b1/plugins/math-proof/LICENSE
Supports: Redistribution terms, Unfilled copyright appendix
https://github.com/anthropics/claude-plugins-official/blob/b78ac49cdc6b3d7b61c4439470e311f4291265b1/plugins/math-proof/README.md
Supports: Whole-plugin installation, Runtime prerequisites, Model usage and resume procedure
https://github.com/anthropics/claude-plugins-official/blob/b78ac49cdc6b3d7b61c4439470e311f4291265b1/plugins/math-proof/skills/siege/SKILL.md
Supports: Agent dependencies, Completion-marker handling, Fresh worker launches and interruption recovery