artagon-vault-mcp
Read-only MCP server for querying an evidence-aware knowledge vault with temporal and provenance-aware data, supporting agent memory and semantic graph projections.
README
Artagon Agentic Temporal Knowledge Vault
An adoption-ready Obsidian repository and deterministic Python metadata compiler for maintaining one evidence-aware knowledge source across agents, humans, semantic graphs, and research packages.
Start here
- Humans: open
00-Home/Start Here.mdin Obsidian. - Agents: read
AGENTS.md, then load the smallest task-specific skill. - Adopters: read
ADOPTION.mdandplugins/artagon-vault-curator/INSTALL.md.
CLI=./plugins/artagon-vault-curator/scripts/artagon-vault
$CLI --vault . status
$CLI --vault . schema-status
$CLI --vault . events-verify
$CLI --vault . compiler-plan --target all
Authoritative layers
immutable originals
canonical Markdown
versioned schema and mappings
append-only semantic event ledger
Everything else is generated from a normalized intermediate representation:
- temporal and provenance-aware JSON/JSONL;
- RDF, JSON-LD, SKOS, PROV-O, OWL-Time, SHACL, CSV, GraphML, and Neo4j projections;
- trust-aware agent memory;
- Google Open Knowledge Format-compatible bundle;
- schema.org web metadata;
- Data Commons observation adapter files;
- RO-Crate research/release metadata;
- Quartz and GitBook human views;
- exact-file manifests and an honestly scoped semantic digest.
Compiler workflow
# After editing approved canonical knowledge:
$CLI --vault . events-reconcile --actor ART-PERSON-OR-AGENT-ID --apply
$CLI --vault . compiler-build --target all --apply
$CLI --vault . compiler-audit --strict
$CLI --vault . publish --apply --strict
Canonical IDs are immutable merge keys. Titles, filenames, paths, URLs, aliases, slugs, and legacy R-numbers are not identities. Staged material and agent proposals remain untrusted until reviewed.
Skills and maintenance agents
The repo-local plugin contains twenty-one Agent Skills and ten specialized maintenance role cards. The roles cover compiler operation, temporal consistency, provenance, graph projection, human publishing, intake, MCP security, store indexing, release integrity, and adversarial review.
Third-party Obsidian, GitBook, Quartz, and LLM-wiki skills are linked under plugins/artagon-vault-curator/references/THIRD_PARTY_SKILLS.md; they are not trusted or installed automatically.
Integrity boundary
File hashes prove exact-byte integrity. The bundled semantic digest applies only to the compiler’s blank-node-free ground N-Quads and is not presented as full RDFC-1.0. Integrity does not establish epistemic truth.
Obsidian, MCP, and stores
The package includes portable Obsidian settings, a prebuilt and reproducible SQLite FTS5 index, loopback-bound examples for Neo4j, PostgreSQL/pgvector, Qdrant, and Fuseki, and a read-only MCP server with client templates for Claude, Codex, VS Code, and generic stdio clients.
pip install -e './plugins/artagon-vault-curator[mcp,stores,compiler]'
artagon-vault --vault . workspace-sync --profile research
artagon-vault --vault . workspace-sync --profile research --apply
artagon-vault --vault . store-build --target sqlite --apply
artagon-vault --vault . integration-audit --strict
artagon-vault-mcp --vault .
Read INTEGRATIONS.md before enabling external services or copying client-specific configuration.
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。