Doc2math
Convert narrative technical documents into grounded Mathematical Problem Specifications with variables, constraints, objectives, and uncertainty.From its SKILL.md
npx -y skills add sinhoneyy/master-skills --skill doc2mathAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 17 stars17 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
3.9 KB, 930 tokens by cl100k_base, as published. Nobody here has run it
DOC2MATH — Document-to-Mathematics Problem Specification
When to Use This Skill
- "Formalize this problem statement into math"
- "Extract the mathematical structure from this research paper section"
- "What variables, constraints, and objectives are in this spec?"
- "Convert this word problem to a structured MPS"
- "Find what's missing in this problem formulation"
Zero-Inference Protocol (Mandatory)
- Closed World — if it is not stated in the document, it does not exist in output
- Grounding Rule — every element must cite the exact source phrase (
"evidence"field) - No Silent Filling — unknown values use
null; ambiguous types use"ambiguous" - Inference Tagging — structural inferences tagged
"inferred": truewith"inference_basis" - MISSING Markers — elements mentioned but insufficiently defined get
"status": "MISSING"with"missing_reason" - No Hallucinated Math — never introduce equations or values not in the source text
Limitations
- Does not invent missing equations, domains, values, or assumptions that are absent from the source document.
- Requires enough source text to cite every extracted element; sparse prompts should be returned with explicit missing-information markers.
- Produces a formal specification, not a solved optimization model or proof.
How It Works
Step 1 — Receive Document
Accept the document text, research excerpt, problem description, or specification as input.
Step 2 — Classify
Identify problem_class: optimization | classification | simulation | proof | estimation | other
Step 3 — Extract MPS Components
Variables — id, name, symbol, type, domain, units, role, evidence, inferred, status
Operators — id, name, symbol, arity, acts_on, produces, evidence, inferred
Constraints — id, type, expression, variables_involved, evidence, hardness, inferred, status
Objectives — id, direction (minimize/maximize/satisfy/find/prove), expression, variables_involved, evidence, inferred
Uncertainty — id, type (stochastic/epistemic/measurement/model/none_stated), affects, characterization, evidence, status
Step 4 — Surface Missing Information
Identify what the document implies but doesn't state: missing_information[] with element, needed_for, missing_reason.
Step 5 — Validate and Score
validation_flags:
has_complete_objectives: true/false/partialhas_bounded_variables: true/false/partialhas_evidence_for_all_elements: true/false/partialinference_count: integermissing_count: integeroverall_formalizability: HIGH/MEDIUM/LOW
Output Format
Produce the complete MPS as a JSON object:
{
"mps_version": "1.0",
"source_title": "...",
"problem_class": "optimization",
"variables": [...],
"operators": [...],
"constraints": [...],
"objectives": [...],
"uncertainty": [...],
"missing_information": [...],
"validation_flags": {
"overall_formalizability": "HIGH"
}
}
Best Practices
- ✅ Apply all 6 Zero-Inference Protocol rules before outputting any element
- ✅ Surface MISSING markers rather than silently inferring — incomplete formalization is valid output
- ✅ Cite the exact source phrase in every
evidencefield - ❌ Never introduce mathematical relationships not grounded in the source text
Additional Resources
- Repository: thebrierfox/doc2math-skill
- Full BYOK tool: ace-license-server-production.up.railway.app/byok/doc2math
- Built by IntuiTek¹ (~K¹) — MIT License
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.