{"id":"bc82a06f-d48b-478a-a26e-508447f87134","slug":"topprismwiki","kind":"skill","name":"topprismwiki","summary":"Evidence-governed knowledge infrastructure for enterprise AI agents, with approved sources, traceable facts, review-gated publication, and agent-ready context.","owner":"topprismdata","repoUrl":"https://github.com/topprismdata/topprismwiki","license":"Apache-2.0","licenseTier":"permissive","trust":"community","stars":3,"match":null,"fewStars":true,"format":"skill","createdAt":"2026-08-20T08:49:53.000Z","addedAt":"2026-10-14T06:22:07.076Z","logoUrl":"https://avatars.githubusercontent.com/u/273794599?u=ddabf7dd2f8f548a1e53cc49347a5e99bca6c099&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["memory-knowledge"],"tags":["agentic-ai","ai-agents","enterprise-ai","human-in-the-loop","knowledge-governance","knowledge-graph","knowledge-provenance"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-26T23:17:31.544Z","sample":false,"url":"https://conduits.dev/tools/topprismwiki","language":"Python","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"

\n \"TopPrism\n

\n\n# TopprismWiki\n\n> **Language / 语言:** English primary · [中文 README](README.zh-CN.md)\n\n\n## Enterprise Knowledge Governance Layer for Agentic AI\n\n[![Python](https://img.shields.io/badge/python-3.11%2B-3776AB.svg)](https://www.python.org/)\n[![CI](https://img.shields.io/badge/CI-GitHub%20Actions-2088FF.svg)](.github/workflows/ci.yml)\n[![License](https://img.shields.io/badge/license-Apache--2.0-green.svg)](LICENSE)\n[![Agent Skill](https://img.shields.io/badge/Agent%20Skill-%24topprismwiki-6E56CF.svg)](skills/topprismwiki/SKILL.md)\n[![TopPrism Native AI](https://img.shields.io/badge/TopPrism-Native%20AI-7C3AED.svg)](https://github.com/topprismdata)\n[![Privacy](https://img.shields.io/badge/data-public%20examples%20only-0B7A75.svg)](docs/en/privacy.md)\n\nTopprismWiki is an evidence-governed knowledge layer that turns approved\nenterprise evidence into traceable formal knowledge for AI agents.\n\nTraditional retrieval helps agents find information. TopprismWiki governs what\ncan become knowledge, why it is trusted, and how it may be published and\nconsumed.\n\nIt is implemented as an Agent Skill package and a dependency-free Python\nreference runner. The long-term direction is an enterprise knowledge operating\nlayer; the public repository currently provides the governed contracts and\nreference workflow needed to build toward that direction.\n\n## Why TopprismWiki\n\nAgentic systems do more than retrieve documents. They summarize, recommend,\nupdate records, and take actions. That makes provenance and change control part\nof the knowledge problem:\n\n- Which source is approved for this decision?\n- What evidence supports a fact or relationship?\n- How should conflicting statements coexist?\n- Who or what may publish a change?\n- Can an agent distinguish formal knowledge from a candidate or an unresolved\n source?\n\nTopprismWiki places a governed publication layer between enterprise sources and\nagent-facing context:\n\n~~~text\nEnterprise evidence\n |\n v\nSource adapters and evidence contracts\n |\n v\nReview, entity and relationship governance\n |\n v\nDeterministic formal publication\n |\n v\nTrusted Knowledge Vault\n |\n v\nQuery, context and graph projections for AI agents\n~~~\n\n## What it is\n\nTopprismWiki provides:\n\n- evidence contracts for approved sources and adapter-produced records;\n- source hashes, evidence anchors, review packages and conflict states;\n- canonical entities and typed relationships with separate candidate overlays;\n- review-gated, atomic publication into a formal Markdown Vault;\n- bounded query, context, graph, batch and diagnostic projections;\n- a single public entry Skill, $topprismwiki, with internal child modules.\n\nTopprismWiki is not a vector database, a general-purpose RAG framework, an\nautomatic truth generator, or a bundled WeChat/Office/OCR connector. The public\nrunner deliberately separates built-in governance behavior from deployment-owned\nadapters and model services.\n\n## Enterprise Knowledge Lifecycle\n\n~~~mermaid\nflowchart TB\n S[Enterprise sources] --> A[Source adapters]\n A --> W[Evidence Workspace]\n W --> G[Review and governance]\n G --> C[Deterministic committer]\n C --> V[Trusted Knowledge Vault]\n V --> R[Query, context and graph]\n R --> AG[AI agents]\n W -. conflicts, candidates, status .-> R\n~~~\n\nThe three-layer boundary is intentional:\n\n| Layer | Role | Write rule |\n| --- | --- | --- |\n| Formal Vault | Approved facts, canonical entities, decisions and typed relationships | Only the deterministic committer writes |\n| Evidence Workspace | Originals, snapshots, hashes, review packages, conflicts, candidates and ledgers | Mutable and rebuildable |\n| Interaction projections | Query, context, graph, status, batch and dashboard views | Read-only projections |\n\nCo-occurrence, group names, member lists, filenames and brand mentions do not\nestablish formal relationships by themselves.\n\n## Core capabilities\n\n| Capability | Public status |\n| --- | --- |\n| Project initialization, JSONL preview and batch ledger | Built-in |\n| Review-gated atomic Markdown publication and rollback | Built-in |\n| Formal query, bounded context and REL graph projection | Built-in |\n| Read-only batch dashboard and diagnostics | Built-in |\n| WeChat, Office, PDF, image and media source processing | Adapter-required |\n| Per-source production watermarks and Obsidian post-commit validation | Not yet enforced by the public runner |\n\nSee the [capability matrix](skills/topprismwiki/references/capability-matrix.md)\nfor the complete boundary.\n\n## 60-second synthetic demo\n\nThe public demo uses synthetic chat events only. It verifies the same governed\npath without exposing private sources:\n\n~~~bash\ngit clone https://github.com/topprismdata/topprismwiki.git\ncd topprismwiki\npython3 -m unittest discover -s skills/topprismwiki/tests -v\n~~~\n\nFor the complete end-to-end path, follow the\n[synthetic walkthrough](docs/en/synthetic-walkthrough.md). A clean run\ncurrently demonstrates:\n\n~~~text\n2 processing units\n11 input messages\n1 formalized fact\n1 evidence-insufficient unit\n1 atomically committed formal page\naccepted post-commit validation\n~~~\n\nPreview creates Workspace evidence and does not write the formal Vault.\nUpdate requires an explicit authorization flag and a reviewed decision package.\n\n## How it differs from common patterns\n\nThis is a comparison of representative architecture patterns, not a benchmark\nor a claim about every product in a category.\n\n| Concern | Document knowledge base | Typical RAG pipeline | Typical GraphRAG pipeline | TopprismWiki |\n| --- | --- | --- | --- | --- |\n| Primary focus | Store and read documents | Retrieve passages for generation | Retrieve and connect graph context | Govern what becomes formal, traceable knowledge |\n| Source approval | Usually external to the store | Application-defined | Application-defined | Explicit source and admission contracts |\n| Evidence anchors | Varies by implementation | Often passage-level | Often document/edge-level | Required for formal facts and operations |\n| Relationship promotion | Usually manual or implicit | Usually outside retrieval | Often extraction-driven | Typed, evidence-backed, candidate-separated |\n| Conflict handling | Often manual | Prompt/application responsibility | Graph/application responsibility | Explicit conflict and unresolved states |\n| Publication control | Tool or process dependent | Pipeline dependent | Pipeline dependent | Review package, hash check and atomic commit |\n| Agent context | Search or document access | Retrieved passages | Graph-plus-text context | Bounded formal facts, relationships and evidence gaps |\n\n## Current evidence and evaluation\n\nThe current public package is validated as a governance reference workflow, not\nas a production-scale knowledge quality benchmark.\n\n- package validation covers eight nested Skill modules;\n- the test suite covers authorization, preview isolation, atomic update,\n rollback, query, graph, retry and diagnostic paths;\n- the public safety scan rejects private paths, credentials and unsanitized\n source examples;\n- the synthetic walkthrough verifies preview, reviewed update, validation,\n query, graph and dashboard behavior.\n\nThe [evaluation guide](docs/en/evaluation.md) defines the next measurement\nlayer: traceability coverage, admission precision, conflict handling,\nidempotence, rollback correctness and, later, labeled knowledge-quality and\nagent-task benchmarks. No accuracy, hallucination-reduction or deployment\nimprovement claim is made by this repository today.\n\n## Quick start\n\nRead [Getting started](docs/en/getting-started.md) and run the synthetic\nwalkthrough before connecting real sources.\n\n~~~bash\npython3 skills/topprismwiki/scripts/topprismwiki.py init --project /path/to/my-wiki\npython3 skills/topprismwiki/scripts/topprismwiki.py doctor --capability core --project /path/to/my-wiki\npython3 skills/topprismwiki/scripts/topprismwiki.py preview --source all --project /path/to/my-wiki\npython3 skills/topprismwiki/scripts/topprismwiki.py status --project /path/to/my-wiki\npython3 skills/topprismwiki/scripts/topprismwiki.py dashboard build --project /path/to/my-wiki\n~~~\n\nIf a command returns blocked or a non-zero exit code, run diagnose first:\n\n~~~bash\npython3 skills/topprismwiki/scripts/topprismwiki.py diagnose --project /path/to/my-wiki\n~~~\n\nOnly an explicitly authorized update with a reviewed decision package may write\nthe formal Vault:\n\n~~~bash\npython3 skills/topprismwiki/scripts/topprismwiki.py update \\\n --source documents --authorized --project /path/to/my-wiki\n~~~\n\n## Documentation\n\n- [Concepts](docs/en/concepts.md)\n- [Product architecture](docs/en/product-architecture.md)\n- [Governance model](docs/en/governance-model.md)\n- [Adapter contracts](docs/en/adapters.md)\n- [Dependencies and agent routing](skills/topprismwiki/references/dependencies-and-routing.md)\n- [Operations and state semantics](docs/en/operations.md)\n- [Troubleshooting](docs/en/troubleshooting.md)\n- [Privacy contract](docs/en/privacy.md)\n- [Evaluation](docs/en/evaluation.md)\n- [Roadmap](ROADMAP.md)\n- [中文文档](README.zh-CN.md)\n\n## Privacy and contribution boundaries\n\nThe repository contains synthetic contacts, chats, customers, projects,\nfilenames, hashes, paths and timestamps only. Real messages, business\ndocuments, Vaults, Workspaces, ledgers, model endpoints and credentials must\nremain outside Git.\n\nRun the public safety scanner before release:\n\n~~~bash\npython3 skills/topprismwiki/scripts/check_public_safety.py .\n~~~\n\nSee [SECURITY.md](SECURITY.md) and [CONTRIBUTING.md](CONTRIBUTING.md) before\nopening an issue or pull request.\n\n## TopPrism status\n\n| Field | Value |\n|---|---|\n| Purpose | Native AI · Evidence-Governed Knowledge Layer |\n| Maturity | Public reference implementation |\n| Evidence | Synthetic walkthrough and repository test suite; no production-scale knowledge-quality benchmark claim |\n| Boundary | Not a vector database, general RAG framework or automatic truth generator |\n| Related | Agent-facing context, formal knowledge publication and source adapters |\n\n## License\n\nApache-2.0. See [LICENSE](LICENSE).\n"}