@phasefolio/mcp
Provides AI agents with access to PhaseFolio's biotech rNPV and probability-of-success engine for asset valuation, supporting public queries and authenticated project management.
README
@phasefolio/mcp
The PhaseFolio MCP client — exposes PhaseFolio's biotech rNPV / probability-of-success engine to any AI agent that speaks the Model Context Protocol.
PhaseFolio is an audit-grade asset valuation workspace for healthcare investors and tech transfer offices. This MCP server gives AI agents direct access to the same engine, scenarios, evidence registers, signed dossier exports, methodology, and competitive landscape that human analysts use in the product.
Install
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"phasefolio": {
"command": "npx",
"args": ["-y", "@phasefolio/mcp"]
}
}
}
For org-scoped tools (your own scenarios, projects, evidence, signed exports), set your bearer token:
{
"mcpServers": {
"phasefolio": {
"command": "npx",
"args": ["-y", "@phasefolio/mcp"],
"env": {
"PHASEFOLIO_TOKEN": "pft_..."
}
}
}
}
Restart Claude Desktop.
Claude Code
HTTP-native — no wrapper needed:
claude mcp add --scope user --transport http phasefolio https://app.phasefolio.com/api/mcp
# with bearer:
claude mcp add --scope user --transport http phasefolio https://app.phasefolio.com/api/mcp \
--header "Authorization: Bearer pft_..."
Codex CLI
codex mcp add phasefolio --http https://app.phasefolio.com/api/mcp
Cursor / other MCP clients
Use the Claude Desktop snippet above (stdio) or point your client at https://app.phasefolio.com/api/mcp if it speaks streamable HTTP.
Tools
Nine tools across two auth tiers.
Public (no token)
query_benchmarks— anonymized network statistics across the PhaseFolio scenario base (PoS by indication × modality, cost / duration percentiles)verify_export— verify a signed PhaseFolio dossier by content hash or URL; returns issued timestamp, methodology version, originating-org identifierget_methodology— fetch any methodology section in citable form (backtest, PoS calibration, IRA framework, evidence standards, network benchmarks)
Bearer (PHASEFOLIO_TOKEN env var, scoped to your org)
get_project— project metadata: indication, sub-indication, modality, biomarker, asset name, stage at entrylist_scenarios— scenarios in a project with top-line eNPV / rNPVget_scenario— full scenario inputs + computed outputs (eNPV, rNPV, cumulative PoS, per-stage breakdown, top sensitivity drivers)get_evidence— evidence-register entries: citations, sources, supporting documentsget_dossier— structured dossier JSON, mirroring the IC Dossier PDF / Excel exportsquery_landscape— asset-anchored competitive landscape (comparable trials, competing programs, sponsor activity, biomarker overlap) sourced from CT.gov + FDA + curated enrichment
Get a bearer token
Bearer tokens are issued per organization. Sign up at app.phasefolio.com (free Research Tier available for academics) — token issuance is currently manual for design partners; contact us at phasefolio.com/contact for access.
Verify a signed export
Every PhaseFolio dossier export carries a cryptographic signature that ties the file content to the engine version and methodology version that produced it. AI agents can verify any PhaseFolio file in one call:
verify_export({ hash: "<sha256-of-the-file>" })
or by URL:
verify_export({ artifact_url: "https://..." })
The public key lives at https://app.phasefolio.com/.well-known/phasefolio-pubkey.pem. Methodology versions and the verify endpoint are documented at https://app.phasefolio.com/methodology.
Links
- Product: https://app.phasefolio.com
- Methodology hub: https://app.phasefolio.com/methodology
- Public verify endpoint: https://app.phasefolio.com/api/v1/verify
- MCP discovery: https://app.phasefolio.com/.well-known/mcp.json
- Official MCP Registry listing: https://registry.modelcontextprotocol.io/v0/servers?search=phasefolio
- Smithery listing: https://smithery.ai/servers/phasefolio/mcp
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。