A
licenseNot graded
qualityB
maintenanceA read-only MCP connector for searching, fetching, and citing provenance-tracked legal corpora with verifiable content hashes.
Apache 2.0
DeepLaw
by Eysn0130
This server provides read-only access to a DeepLaw Knowledge Asset vault, allowing agents to search, retrieve, verify, and compile bounded knowledge capsules from human-reviewed, active knowledge assets. It is strictly non-destructive: no writes, mutations, learning, or access to case data.
Search: Perform semantic and lexical searches over knowledge assets using text queries, goals, tasks, knowledge kinds, and memory tiers. Returns ranked result cards with excerpts, metadata, tags, trust level, and source references. (Numeric confidence scores are not exposed.)
Get: Retrieve the full content of a specific asset by its unique ID, including its statement, source references, directive mode, sensitivity, provenance metadata, and integrity information.
Context: Compile a bounded Knowledge Capsule for a given task (and optional goal), aggregating relevant constraints, decisions, assets, experiences, open questions, relations, evidence, gaps, and next actions. Respects character/item budget limits and requires confirmation that no case data is included.
Verify: Validate the integrity and authenticity of an asset by checking its content hash, audit chain, source file bindings, and agent-usability.
Inspect: Obtain an overview of the vault’s health and readiness, including asset/source/fragment/relation counts, status breakdowns (active, proposed, quarantined, etc.), memory tier and kind distributions, audit integrity, and recommended next actions.
Key constraints: Only human-verified, active assets are surfaced. The server is read-only, idempotent, and non-destructive—it cannot remember, learn, approve, import, mutate, or access case data. Knowledge assets are explicitly not legal authority; use the separate law_support tool for official legal sources.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@DeepLaw查找刑法第二百六十六条的具体内容"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
DeepLaw 将原始资料编译为受治理的知识与 Living Wiki，并为当前任务返回有界、可验证的 Knowledge Capsule。它不是普通 RAG、完整 transcript 仓库、Obsidian 替代品、法律裁判系统或 Agent runtime。
架构冻结为一个共享治理内核上的三个产品角色：
Task Continuity / Governed Project Knowledge；
Source-native Evidence Library；
Living Wiki。
三者共用一个 Context Compiler：
Discovery → Admission → Selection → Bounded Verifiable Knowledge Capsule。Context Compiler
不是第四产品或第二检索引擎；Legal Pack 是 Evidence Library 的第一方法律策略面。专业来源保留
原始字节、版本、Fragment 和 Locator，Wiki 是可重建投影，不是完整可编辑 canonical 副本。
当前诚实状态
公开 package/main：
0.12.0 Beta；最新 tag：v0.12.0。Active qualification：
machine_evaluation_pending，profile：kernel_release_core，Gate classification：v9。release_ready=false，尚无0.13.0tag 或 release。当前 Provider advertisement：knowledge-support input v7 / output v6，仅
query、context、explain；input v1-v6 和 output v1-v5 仅为 compatibility/internal。本地 regression、mock、dry-run、旧报告或 no-model smoke 不构成真实 Host、Human Gold、法律专家、 3 OS、scale、供应链或发布证据。Kernel Release Core、Capability 与 Competitive/Research Claim 分别判定；缺失的可选能力或研究证据保持
not_executed，只禁止对应声明。v0.13 Kernel 的支持上限为每个 Vault 10,000 个 active governed Knowledge Objects；>10k 为实验范围，100k 与其 sharding/bundling 不属于 v0.13。官方 signed Legal Pack、GUI/Desktop interoperability、semantic restore 与 Claude 未经各自证据不得宣称发布。
机器状态只读取
benchmarks/v013/active-qualification-v3.json 和
benchmarks/release/v013-gate-classification-v9.json；
README 不承担第二状态台账。
Related MCP server: Jusratio Case File
安装
正式版本使用 Python 3.11+ 与 uv：
uv tool install \
https://github.com/Eysn0130/DeepLaw/releases/download/v0.12.0/deeplaw-0.12.0-py3-none-any.whl
deeplaw --version仓库开发环境：
uv sync --all-extras首要产品旅程
先建立并检查本地 Vault。doctor 必须报告 canonical/autonomous readiness；缺失前置条件时应返回
可操作 Gap，而不是继续连接 Host。
deeplaw knowledge init --vault ./vault --name my-project --scope project
deeplaw knowledge doctor --vault ./vault仓库开发环境提供一个可复制、公开、source-free、无模型的最短成功流程。它在新目录中执行 Source add、owner source review、只读 Host handoff、现有 Coordinator/MCP grant 编译、Query、 Context，以及 Wiki 到 exact Source Revision 的下钻，并在 JSON 中逐项报告结果：
uv run python -m examples.living_wiki.run_demo \
--workspace /tmp/deeplaw-living-wiki-demo该流程是本地 development evidence，不是真实 Host、外部 benchmark、qualification 或 release
evidence。完整的逐步 CLI/Host packet 工作流见
docs/LIVING_WIKI_COMPILER.md。
正式 MCP 配置必须使用闭合环境入口；以下命令用于 owner 诊断，静态 Host 配置由后续
host connect 生成同一 argv：
deeplaw knowledge mcp --closed-environment --stdio
deeplaw mcp --closed-environment --stdio建立任务线并生成 task-neutral、只读、人工合并的 Host 配置。静态 host connect 不选择 task，
不管理 Host 登录或 runtime，也不启用 knowledge_sink。
deeplaw knowledge task start --vault ./vault \
--project DeepLaw --task 'Finish the selected task.' --workspace .
deeplaw knowledge task locate --vault ./vault \
--project DeepLaw --task 'Finish the selected task.' --workspace .
deeplaw knowledge host connect --host codex --vault ./vault首次 session 绑定是显式 owner mutation，要求现有 Sink grant、幂等键和当前 workspace。首选入口只从
stdin 读取一次 raw official session ID，并立即绑定它的 SHA-256；raw ID 不得出现在 argv、Ledger、
日志、receipt 或 Provider。bind-host-session 仅保留给已经由 owner 安全计算 SHA-256 的调用方。
deeplaw knowledge sink enable --vault ./vault \
--writer-id owner-host-continuity --scope project --max-sensitivity private \
--operation record_run --operation remember --operation forget
deeplaw knowledge task enroll-host-session --vault ./vault \
--host codex \
--task-handle TASK_HANDLE --workspace . --grant-id GRANT_ID \
--idempotency-key BIND_IDEMPOTENCY_KEY --confirm-no-case-data \
< OWNER_ONLY_OFFICIAL_SESSION_ID
deeplaw knowledge task checkpoint --vault ./vault \
--task-handle TASK_HANDLE --workspace . --grant-id GRANT_ID \
--idempotency-key CHECKPOINT_IDEMPOTENCY_KEY \
--summary 'Bounded verified progress.' --next-action 'Continue the selected task.' \
--expires-at '2099-01-01T00:00:00Z' --confirm-no-case-data
deeplaw knowledge task resolve-host-continuity --vault ./vault \
--host codex --session-sha256 SESSION_SHA256_FROM_ENROLLMENT_RESULT --workspace .
deeplaw knowledge task resume --vault ./vault \
--project DeepLaw --task 'Finish the selected task.' --workspace .
deeplaw knowledge task timeline --vault ./vault \
--task-handle TASK_HANDLE --workspace .普通恢复不要求 task handle。fork、compaction、stale checkpoint、wrong task/worktree、ambiguous binding 和 selective forget 都必须重新校验当前状态并 fail closed 为结构化 Gap。
只导入来源尚未产生可 Admission 的编译知识。此时 context 必须返回
uncompiled_source Gap；不能把它描述为成功旅程。完成上述 source review、handoff、grant 和现有
Coordinator 编译后，再通过同一 Context Compiler 获得有界上下文；精确引用任务按需下钻到
Source Revision、Fragment 和 Locator，而不是把整个来源复制进 Wiki 或 Provider。
deeplaw knowledge source add --vault ./vault --source ./guide.md \
--confirm-no-case-data
deeplaw knowledge compile handoff --vault ./vault \
--source-revision-id sourcerev_REPLACE固定边界
自主写入不等于权威升级。embedding、图权重、模型 confidence、引用次数或使用频率都不能产生 官方身份、法律 Authority 或法律适用结论。
knowledge_support永久只读；knowledge_sink是独立、显式 grant 控制的写进程；law_support独立只读。不自动读取或保存 prompt、transcript、hidden reasoning、auth、Secret 或 raw log。
Provider 不得收到路径、session hash、内部 selection/receipt identity、未 admitted 内容或 Secret。
普通 read 不写 canonical Ledger；只有 durable mutation 进入共享 Coordinator。
不增加远程 canonical storage、telemetry、GUI/云控制平面或新知识引擎。
完整规范与子系统导航见 docs/README.md，安全边界见
SECURITY.md。DeepLaw 采用 Apache License 2.0。
Available Tools
1 toolknowledge_supportARead-onlyIdempotent
Optional read-only gateway for an explicitly selected DeepLaw Knowledge Asset vault. It searches only human-reviewed active assets and compiles bounded task capsules. It cannot remember, learn, approve, import, mutate, or access Analytix case projects.
ParametersJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| goal | No | ||
| task | No | ||
| kinds | No | ||
| limit | No | ||
| query | No | ||
| asset_id | No | ||
| max_chars | No | ||
| operation | Yes | search | |
| memory_tiers | No | ||
| confirm_no_case_data | No |
Output Schema
ParametersJSON Schema
| Name | Required | Description |
|---|---|---|
| result | Yes | |
| operation | Yes | |
| schema_version | Yes | |
| authority_boundary | Yes |
TDQS
A4/5.0
Behavior5/5
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Despite annotations already declaring readOnlyHint, openWorldHint=false, idempotentHint, and no destructive behavior, the description adds substantial behavioral context: it clarifies that only 'human-reviewed active assets' are accessible, that outputs are 'bounded task capsules', and that it explicitly cannot perform other operations. This goes beyond the annotations and informs the agent of critical limitations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Conciseness5/5
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, front-loaded with the core purpose, and every word adds value. It efficiently conveys the tool's nature, limitations, and scope without unnecessary elaboration.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Completeness3/5
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description gives a strong high-level overview and constraints, but for a tool with 10 parameters and 5 operations, it doesn't explain the distinct behaviors of each operation (search, get, context, verify, inspect). The output schema exists, so return values are covered, but the operational semantics are under-specified. It is adequate but with clear gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Parameters1/5
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has 0% description coverage for 10 parameters, and the description does not explain any of the parameters (operation, query, asset_id, etc.). With such low coverage, the description was expected to compensate, but it offers no parameter-level guidance, leaving the agent to infer meaning solely from parameter names and enum values.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Purpose5/5
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states a specific verb+resource: it is a 'read-only gateway' for a 'DeepLaw Knowledge Asset vault'. It precisely defines scope with constraints like 'searches only human-reviewed active assets' and 'compiles bounded task capsules', which distinguishes it from potential sibling tools by also listing what it cannot do.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Usage Guidelines4/5
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives clear context for usage: it is 'optional' and 'explicitly selected', implying it is used when interacting with a specific vault. It provides exclusion boundaries ('cannot remember, learn, approve, import, mutate, or access Analytix case projects'), effectively telling when not to use this tool, though it doesn't name explicit alternatives.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
1 tool update
v0.11.0- First observed
knowledge_support
TDQS
A3.9/5.0
Disambiguation5/5
With only one tool, there is no possibility of confusion between tools. The tool's description clearly defines its read-only scope, making its purpose unambiguous to an agent.
Naming Consistency5/5
The sole tool uses a clear snake_case name ('knowledge_support'), which follows a consistent naming style. With no other tools to compare against, there is no inconsistency.
Tool Count3/5
A single tool feels thin, placing it at the borderline of acceptable scope. While it may suffice for a narrow read-only knowledge search, the server's name suggests a broader legal domain that is not reflected here.
Completeness2/5
The tool is strictly read-only and explicitly lacks capabilities such as create, update, delete, import, or access to case projects. For a knowledge-support server, the absence of any management or integration functions represents a significant gap that could hinder agents.
Maintenance
ActivityActive
ResponsivenessNo issues
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Connectors
Read-only Remote MCP for externally grounded AI agent trust receipts.
Read-only MCP server for The Quiet Protocol's engines, benchmarks, proof, and business data.
A paid remote MCP for Statewright, built to return verdicts, receipts, usage logs, and audit-ready J
A paid remote MCP for CodeG, built to return verdicts, receipts, usage logs, and audit-ready JSON.
Related MCP Servers
- AlicenseBqualityAmaintenanceLocal-first MCP server that transforms civil case PDFs into queryable cases with structural provenance, evidence IDs, and page-level verification for drafting and factual review.22Apache 2.0
- FlicenseAqualityBmaintenanceMCP server that provides access to Argentine legal documents (legislation, CSJN jurisprudence, international treaties) with verifiable provenance including SHA256 hashes and source URLs, enabling legal professionals to search and verify citations.13-
- FlicenseNot gradedqualityCmaintenanceRemote MCP server exposing the CounselStack citation engine as nine legal research tools, enabling read-only citation lookup, statute search, and case text retrieval through Claude.-
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Eysn0130/DeepLaw'
If you have feedback or need assistance with the MCP directory API, please join our Discord server