PaperGraph MCP v0.10.0
PaperGraph MCP v0.10.0
v0.10.0 adds Paper Map, an evidence-first first-load overview for a stored paper.
Paper Map helps a researcher or agent quickly see the shape of a paper before choosing where to read next. It highlights likely main-result candidates, local result structure, proof-path evidence, external reading risks, and evidence-quality warnings while preserving PaperGraph's core rule: show evidence, do not guess mathematical dependencies.
Highlights
- Added
workspace_get_paper_map - Added
Workspace.get_paper_map(...) - Added CLI command
get-paper-map - Surfaces main-result candidates with explicit heuristic reasons
- Builds a bounded reading route from existing proof-path evidence
- Embeds external import risks from the existing import planner
- Reports evidence-quality warnings for missing proofs, unresolved references, and external dependencies
- Updated README, onboarding instructions, diagnostics, CI smoke checks, and version pins to
v0.10.0
Evidence Boundaries
Paper Map does not verify proofs, perform semantic theorem matching, infer hidden prerequisites, or automatically import external papers. Empty dependencies still mean “no supported extraction evidence found,” not “there are no mathematical dependencies.”
Testing
434 passed, 1 skipped