name	math-paper-reading
description	Route mathematical paper-reading tasks through structured extraction, dependency analysis, deep reading, literature search, and reference management modules.

Math Paper Reading

Use this skill when a coding agent needs to read, structure, explain, trace, or manage mathematical papers with a data-driven workflow.

Start with agent_router.md. It is the master router and decides which package-local module to load:

skill_base.md: extract definitions, lemmas, theorems, proofs, and dependency structure into JSON.
skill_pathway_proof.md: compute dependency paths and render proof graphs.
skill_paper_deep_read.md: provide layered paper reading and gap filling.
skill_reference_manager.md: maintain the local paper reference database.
skill_literature_search.md: generate targeted literature-search strategies.

Keep fact extraction separate from interpretation.
Prefer structured JSON evidence before graph, reading, or reference updates.
Use package-local examples only as examples; do not treat them as user data.
For complex dependency or merge tasks, write and run scripts instead of estimating manually.

Prefer the router and package-local module files over this thin compatibility entrypoint when executing a concrete workflow.

name	math-paper-reading
description	Route mathematical paper-reading tasks through structured extraction, dependency analysis, deep reading, literature search, and reference management modules.

Use this skill when a coding agent needs to read, structure, explain, trace, or manage mathematical papers with a data-driven workflow.

Start with agent_router.md. It is the master router and decides which package-local module to load:

skill_base.md: extract definitions, lemmas, theorems, proofs, and dependency structure into JSON.
skill_pathway_proof.md: compute dependency paths and render proof graphs.
skill_paper_deep_read.md: provide layered paper reading and gap filling.
skill_reference_manager.md: maintain the local paper reference database.
skill_literature_search.md: generate targeted literature-search strategies.

Keep fact extraction separate from interpretation.
Prefer structured JSON evidence before graph, reading, or reference updates.
Use package-local examples only as examples; do not treat them as user data.
For complex dependency or merge tasks, write and run scripts instead of estimating manually.

Prefer the router and package-local module files over this thin compatibility entrypoint when executing a concrete workflow.

math-paper-reading