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., "@muninn-local-mcp search project memory for dependency management decisions"
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.
A local-first Model Context Protocol (MCP) server that gives AI agents (such as OpenCode) persistent, project-scoped memory powered by ChromaDB and Ollama embeddings.
All data stays on your machine — no external API calls, no cloud storage.
Note: This project is adapted from muninn-mcp and modified to run fully locally using Ollama + ChromaDB instead of cloud services.
Persistent memory — store and recall context across sessions via vector search
Project isolation — each git project gets its own memory namespace, automatically
Global memory — share cross-project knowledge (tooling, patterns, decisions)
Local embeddings — vectors generated by a local Ollama model, zero data leaves your machine
MCP-native — works with any MCP-compatible client (OpenCode, Claude Desktop, etc.)
Zero-config defaults — sensible defaults that work out of the box
Related MCP server: knitbrain
ollama pull mxbai-embed-largegit clone https://github.com/wolfcao/muninn-local-mcp.git
cd muninn-local-mcp
uv syncVariable | Default | Description |
|
| ChromaDB data directory |
|
| Ollama service URL |
|
| Embedding model name |
| (auto from git root) | Force a specific project ID |
Add the server to your opencode.json:
{
"mcp": {
"muninn": {
"type": "stdio",
"command": "uv",
"args": [
"--directory",
"/path/to/muninn-local-mcp",
"run",
"python",
"-m",
"muninn_local"
]
}
}
}Run the MCP server directly:
python -m muninn_localMuninn exposes 7 tools, split between project-scoped and global memory.
Tool | Parameters | Description |
|
| Store a project-scoped memory |
|
| Semantic search within current project |
|
| List memories (newest first) |
|
| Delete a specific memory |
Tool | Parameters | Description |
|
| Store a cross-project memory |
|
| Semantic search across all projects |
|
| List global memories (newest first) |
The memory_type parameter accepts: summary, decision, next-steps, code-pattern, note (default).
Muninn automatically identifies the current project by resolving the git repository root (git rev-parse --show-toplevel) and hashing the path with SHA256. The resulting fingerprint becomes the project_id, ensuring each project's memories are isolated in their own ChromaDB collection.
Layer | Module | Responsibility |
Entry |
| MCP FastMCP server |
Business |
| Memory CRUD operations |
Storage |
| ChromaDB persistence wrapper |
Embedding |
| Vector generation via Ollama API |
Config |
| data_dir / ollama_url / embed_model |
Identity |
| git root → auto project_id |
Ollama must be running — the server depends on a local Ollama instance for embedding generation.
Data is persistent — ChromaDB stores data in ~/.config/opencode/muninn/chroma/. Deleting this directory wipes all memories.
Path-sensitive project IDs — cloning or forking to a different path generates a new project_id, so memories won't carry over. Override with MUNINN_PROJECT_ID if needed.
MIT
List globally stored memories, ordered by creation time (newest first).
Args: limit: Maximum number of memories to return (default 20).
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations are missing, so description carries full burden. It mentions ordering and default limit, but lacks info on safety (read-only), return format (though output schema exists), and any pagination or auth requirements.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise: one sentence and a brief parameter note, with no wasted words. It front-loads the main action.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the presence of an output schema, the description need not detail return values. It adequately covers purpose and parameter, but lacks explicit guidance and safety context for a tool with no annotations.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
With 0% schema coverage, the description compensates by explaining the limit parameter's meaning and default. This adds value beyond the raw schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool lists globally stored memories with a specific ordering (newest first), distinguishing it from siblings like global_memory_search and memory_list.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives such as global_memory_search or memory_list. The description only states what it does without context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the global (cross-project) memory store using semantic similarity.
Embeds the query and returns the most relevant stored memories from all projects.
Args: query: The search query text (required). top_k: Maximum number of results to return (default 5).
| Name | Required | Description | Default |
|---|---|---|---|
| query | Yes | ||
| top_k | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description bears full burden. It discloses semantic similarity and result retrieval but lacks details on embedding model, pagination, max top_k limits, or handling of no results.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Description is concise with an Args section, no extraneous content, and front-loaded purpose. Each sentence serves a purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given output schema exists, description needn't detail return format. It covers purpose and parameters adequately for a search tool, though could mention memory content scope.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, so description compensates. It explains 'query' as required search text and 'top_k' as max results with default 5, adding meaning beyond schema titles.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool searches the global memory store using semantic similarity, specifying the verb (search), resource (global memory), and method (semantic similarity). This distinguishes it from sibling tools like global_memory_write and memory_search.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage for cross-project search but does not explicitly state when to use vs. alternatives like memory_search (project-specific). No exclusions or when-not guidance provided.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Write a memory entry to the global (cross-project) memory store.
Global memories are shared across all projects and can be queried regardless of which project the MCP server is running in.
Args: text: The memory content to store (required). memory_type: Type of memory. Must be one of: summary, decision, next-steps, code-pattern, note (default). tags: Comma-separated tags for categorization (optional).
| Name | Required | Description | Default |
|---|---|---|---|
| tags | No | ||
| text | Yes | ||
| memory_type | No | note |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, and the description does not disclose behavioral traits such as idempotency, overwrite behavior, size limits, or auth requirements. The description is functional but lacks depth.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured with a clear main purpose and bullet points for parameters. It is slightly verbose but efficient overall.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given that an output schema exists, return values are handled. The description covers param semantics and purpose well but lacks behavioral transparency, limiting completeness for a mutation tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description covers all three parameters: text (required, content), memory_type (allowed values with default), and tags (optional, comma-separated). However, there is a mismatch: schema lacks enum enforcement, but description lists values.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool writes a memory entry to the global memory store, specifying it is cross-project. This differentiates it from per-project memory tools.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It explains that global memories are shared across projects, implying when to use this vs memory_write. However, it does not explicitly list alternatives or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Delete a specific memory from the current project by its ID.
Args: memory_id: The full or partial ID of the memory to delete (required).
| Name | Required | Description | Default |
|---|---|---|---|
| memory_id | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries full burden for behavioral disclosure. It only states the deletion action, but does not mention irreversibility, confirmation steps, cascading effects, or any other behavioral traits. This is minimal transparency for a destructive operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise: two sentences, the first stating the purpose and the second explaining the argument. No unnecessary words, front-loaded with the core action.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple one-parameter deletion tool, the description covers the basic purpose and parameter meaning. However, it does not mention the output schema (which exists) or any error conditions, leaving some context gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 0% description coverage; the description adds the nuance that memory_id can be a full or partial ID, which is not in the schema. However, it does not explain the format or how to obtain a memory ID. This partially compensates but lacks completeness.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action (delete), the resource (memory), and the scope (from the current project by its ID). This effectively distinguishes it from sibling tools that list, search, or write memories, including global variants.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description does not provide any guidance on when to use this tool versus alternatives. It mentions 'current project' but does not contrast with global memory tools, and offers no conditions, prerequisites, or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List memories stored in the current project, ordered by creation time (newest first).
Args: limit: Maximum number of memories to return (default 20). offset: Number of memories to skip for pagination (default 0).
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| offset | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description explains ordering and pagination but does not mention safety (read-only implied), output format, or potential rate limits. Adequate but minimal for a read tool.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two concise sentences plus a two-line arg list. No wasted words, main purpose front-loaded. Perfectly sized for a simple list tool.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple list with two optional params and an output schema, the description covers core functionality: listing, ordering, pagination. Lacks mention of filter vs unfiltered listing but sufficient given tool simplicity.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema has 0% description coverage; the description compensates by explaining limit ('maximum number to return, default 20') and offset ('skip for pagination, default 0'). Adds clear meaning beyond schema types and defaults.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states it lists memories in the current project, ordered by creation time newest first. This distinguishes it from sibling tools like global_memory_list (different scope) and memory_search (filtered search).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Implies scope ('current project') but does not explicitly say when to use this tool over alternatives like global_memory_list or memory_search. No guidance on prerequisites or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the current project's memory store using semantic similarity.
Embeds the query and returns the most relevant stored memories.
Args: query: The search query text (required). top_k: Maximum number of results to return (default 5).
| Name | Required | Description | Default |
|---|---|---|---|
| query | Yes | ||
| top_k | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It states the tool performs semantic similarity search and returns relevant memories, but does not disclose any behavioral traits such as read-only nature, auth needs, or edge cases. Adequate but not detailed.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise, with a clear first sentence stating purpose, followed by a brief implementation note, and an ordered list of parameters. No unnecessary information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple search tool with two parameters and an output schema, the description covers purpose, parameters, and method. It does not detail return values, but with output schema present, that is acceptable. Minor gap in behavioral transparency prevents a 5.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The Args section in the description clearly explains both parameters: 'query' as required search text and 'top_k' as maximum results with a default of 5. This adds meaning beyond the schema, which only defines types and default.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that the tool searches the current project's memory store using semantic similarity, distinguishing it from global memory search tools. It specifies the verb 'search' and the resource 'current project's memory store'.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies local scope ('current project') versus global siblings, but does not explicitly state when to use this tool over alternatives like global_memory_search. It provides clear context but lacks exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Write a memory entry to the current project's memory store.
Use this tool to save important context, decisions, code patterns, or next-steps that should be persisted for future sessions.
Args: text: The memory content to store (required). memory_type: Type of memory. Must be one of: summary, decision, next-steps, code-pattern, note (default). tags: Comma-separated tags for categorization (optional).
| Name | Required | Description | Default |
|---|---|---|---|
| tags | No | ||
| text | Yes | ||
| memory_type | No | note |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description bears full responsibility for behavioral clues. It indicates that memories are 'persisted for future sessions,' implying durability. However, it lacks details on whether writes are append-only or overwriting, authorization requirements, size limits, or return values. This leaves room for ambiguity.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise, starting with the core purpose and then detailing parameters. The Args list improves readability. Every sentence is informative, though the parameter descriptions could be slightly more compact without losing clarity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Although an output schema exists, the description does not mention what the tool returns (e.g., success message, memory ID). It also lacks comparison to global_memory_write, which shares similar purpose but with broader scope. The description is adequate for basic use but misses context that would help the agent choose between siblings.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description fully compensates for 0% schema coverage by providing an explicit Args section that explains each parameter: 'text' required, 'memory_type' with valid options (summary, decision, next-steps, code-pattern, note, default), and 'tags' as comma-separated. This adds significant meaning beyond the schema's type/default fields.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: 'Write a memory entry to the current project's memory store.' It specifies the action (write), the resource (memory entry), and scope (current project). This distinguishes it from siblings like global_memory_write, though not explicitly. The examples of memory types further clarify the domain.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides explicit context for when to use the tool: 'Use this tool to save important context, decisions, code patterns, or next-steps that should be persisted for future sessions.' This guides the agent on appropriate usage, though it does not mention when not to use it or explicitly reference alternatives like global_memory_write.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
v0.1.0global_memory_listglobal_memory_searchglobal_memory_writememory_deletememory_listmemory_searchmemory_writeScored across 7 tools
Tools are clearly separated by scope (global vs. project-specific) using consistent prefixes, and each tool performs a distinct operation (list, search, write, delete). No ambiguity between tools.
All tool names follow a consistent pattern: scope prefix (global_memory_ or memory_) followed by a verb (list, search, write, delete). Naming is predictable and uniform.
With 7 tools covering both global and project memory operations (list, search, write, delete), the count is well-scoped and appropriate for a focused memory store server.
The tool surface is mostly complete for both scopes, but the global scope lacks a delete operation, which is a minor gap. CRUD operations are otherwise covered.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Cloud-hosted MCP server for durable AI memory
An MCP server that gives your AI access to the source code and docs of all public github repos
Hosted MCP server connecting claude.ai, ChatGPT and other AI apps to your own computer
An MCP memory server. One memory your agents share — across models, devices and apps.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/wolfcao/muninn-local-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
