Convert research papers and associated files, research code, or both into a reviewed paper skill and tested MCP server. Use for paper-only, code-only, or combined conversions and resumes.
Installs into .claude/skills of the current project.
Are you the author of Paper2agent?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/ahmadhammoud56-paper2agent)
---
name: paper2agent
description: Convert research papers and associated files, research code, or both into a reviewed paper skill and tested MCP server. Use for paper-only, code-only, or combined conversions and resumes.
---
# Paper2Agent
Route by the supplied inputs and requested outcome:
- For paper PDFs and associated files, follow [Paper2Skill](paper2skill/SKILL.md).
- For a research code repository, follow [Paper2MCP](paper2mcp/SKILL.md).
- When both outputs are requested, follow both workflows.
Resolve each component's scripts and references relative to its own directory; bind `SKILL_ROOT` to that component. Keep their work directories separate. Independent paper and code tasks may run concurrently within host limits. Every MCP tool must bind to existing repository code.
Combined output:
```text
dist/<project>-agent/
├── skill/<paper-name>/
└── mcp/<repository-name>-mcp/
```
Build the paper skill at its final component path and pass its strict verification. Complete Paper2MCP's existing runtime and ZIP delivery checks, then stage the verified archive contents under `mcp/` and confirm the files are unchanged. Keep review and development evidence outside the delivery.
Single-mode runs retain their component's output contract. Report delivery paths, verification results and limitations. If either component is blocked, report the combined result as partial.
For questions about Paper2Agent's methods, results, figures, or supplementary material, read the [Paper2Agent paper](paper2agent-paper/SKILL.md).
