de en es ja ko ru zh

mcp-intelligence-context

Name: mcp-intelligence-context
Author: LeoChimal09

by LeoChimal09

Overview Schema Source Related Servers Discussions Admin

Python

Local

Contexto de Inteligencia MCP

Un servidor MCP de Inteligencia de Repositorio que indexa los archivos, símbolos, importaciones y el grafo de dependencias de un código base, y entrega a Copilot/agentes un paquete de contexto pequeño y enfocado en lugar de hacer que escaneen todo el repositorio.

Por qué

Cuando un agente recibe una pregunta ambigua sobre un repositorio grande, a menudo tiene que listar directorios repetidamente, abrir archivos no relacionados y volver a derivar la estructura antes de encontrar el código relevante — quemando tokens y tiempo. Este proyecto construye un índice persistente y actualizado incrementalmente del repositorio (archivos, símbolos, importaciones, dependencias inversas) y expone herramientas MCP que devuelven solo el contexto relevante para una consulta, con un presupuesto aproximado de tokens.

Related MCP server: lens

Cómo funciona

index_repository recorre el repositorio (respetando .gitignore), analiza archivos Python (mediante ast) y JS/TS (mediante heurísticas ligeras de regex) en busca de funciones/clases/métodos/importaciones/exportaciones, y construye un grafo de dependencias inversas. El índice se almacena en caché en .mcp_intel_cache/index.json y se actualiza incrementalmente (solo se vuelven a analizar los archivos modificados, según mtime/tamaño).
search_code / get_relevant_context clasifican archivos por nombre de símbolo, nombre de archivo, docstring/resumen y coincidencias de importación (búsqueda léxica/de símbolos — sin embeddings en este MVP) y devuelven un paquete de contexto con presupuesto de tokens: tablas de símbolos + pequeños extractos de código, no archivos completos. get_relevant_context también informa una comparación de token_savings frente a una línea base ingenua de escaneo completo del repositorio, para que los ahorros sean visibles en la propia respuesta de la herramienta.
get_file_summary / get_dependencies permiten a un agente profundizar en los símbolos de un archivo específico o en el radio de impacto (importadores/importaciones) sin leer el archivo completo.
Las herramientas informan una advertencia de obsolescencia si el índice en caché tiene más de 5 minutos y no hay un observador en vivo activo. En la práctica, la primera llamada a una herramienta para un repositorio inicia un observador de archivos en segundo plano (mediante watchdog) que aplica eventos de creación/modificación/eliminación al índice en memoria inmediatamente, por lo que el índice se mantiene continuamente actualizado a medida que el código cambia — sin necesidad de reindexar manualmente durante una sesión. La caché en disco se vacía con un debounce (~2s) para que los guardados rápidos no provoquen una escritura por pulsación de tecla.

Estructura del repositorio

src/mcp_intelligence_context/   Python MCP server package
  walker.py                     gitignore-aware file walker
  parsers/                      Python (ast) and JS/TS (regex) symbol extraction
  indexer.py                    builds/caches the RepoIndex, resolves imports
  watcher.py                    background file watcher that keeps the index live
  search.py                     lexical/symbol search + reverse-dep lookups
  context_builder.py            token-budgeted context package assembly
  server.py                     MCP tool definitions (stdio server)
vscode-extension/                VS Code extension wrapper (setup/reindex/status commands)
scripts/                         one-command bootstrap for new users

Inicio rápido (Nuevos usuarios)

Si eres nuevo en MCP y solo quieres que esto funcione rápidamente en VS Code:

git clone https://github.com/LeoChimal09/MCP-INTELLIGENCE-CONTEXT.git
cd MCP-INTELLIGENCE-CONTEXT
bash scripts/setup_mcp_workspace.sh

Lo que hace este script:

Instala (o actualiza) mcp-intelligence-context con pipx.
Escribe .vscode/mcp.json para este espacio de trabajo.
Restringe la indexación a la carpeta del espacio de trabajo actual estableciendo MCP_INTEL_ALLOWED_ROOTS=${workspaceFolder}.

Luego en VS Code:

Paleta de comandos -> MCP: List Servers.
Inicia/Reinicia mcp-intelligence-context.
En el selector de herramientas de Copilot Chat, habilita mcp-intelligence-context.

Si el script dice que falta pipx, instálalo una vez:

brew install pipx
pipx ensurepath

Ejecutar el servidor MCP de forma independiente

python3 -m venv .venv
.venv/bin/pip install -e .
.venv/bin/mcp-intelligence-context        # or: python -m mcp_intelligence_context.server

Apunta el repositorio a indexar estableciendo MCP_INTEL_REPO_ROOT, o pasa repo_root explícitamente a cualquier llamada de herramienta (por defecto, el directorio de trabajo actual del servidor).

Instalación sin clonar este repositorio

Otros usuarios no necesitan una copia local — instala directamente desde el repositorio git (o desde PyPI, una vez publicado allí):

python3 -m venv .venv
.venv/bin/pip install "git+https://github.com/LeoChimal09/MCP-INTELLIGENCE-CONTEXT.git"
# once published: .venv/bin/pip install mcp-intelligence-context

El script de consola mcp-intelligence-context y la variable de entorno MCP_INTEL_REPO_ROOT funcionan exactamente igual en ambos casos — solo difiere la fuente de pip install.

Registrar con un cliente MCP (por ejemplo, VS Code)

Añade a .vscode/mcp.json en el espacio de trabajo de destino:

{
  "servers": {
    "mcp-intelligence-context": {
      "type": "stdio",
      "command": "/absolute/path/to/.venv/bin/python",
      "args": ["-m", "mcp_intelligence_context.server"],
      "env": { "MCP_INTEL_REPO_ROOT": "${workspaceFolder}" }
    }
  }
}

Extensión de VS Code

vscode-extension/ incluye un envoltorio ligero con tres comandos:

MCP Intelligence: Setup Server — crea un venv e instala el paquete de Python, luego escribe la entrada .vscode/mcp.json anterior.
MCP Intelligence: Reindex Repository — fuerza una reindexación del espacio de trabajo abierto.
MCP Intelligence: Show Status — imprime el recuento de archivos del índice en caché, el commit de git y la antigüedad.

Por defecto, "Setup Server" instala el paquete desde el repositorio git de este proyecto en un venv bajo el almacenamiento privado de la extensión — sin necesidad de clon local. Dos ajustes controlan esto:

mcpIntelligenceContext.serverPath — apunta a una copia editable local (utilizada para el desarrollo en este monorepo); déjalo vacío en caso contrario.
mcpIntelligenceContext.pythonPackageSource — anula el destino de instalación de pip (por ejemplo, un nombre de paquete de PyPI) cuando serverPath está vacío.

Para compilarlo:

cd vscode-extension
npm install
npm run compile

Luego presiona F5 en VS Code (con vscode-extension/ abierto) para lanzar un Host de Desarrollo de Extensiones.

Herramientas MCP disponibles

Herramienta	Propósito
`index_repository`	Construir/actualizar el índice para una raíz de repositorio
`get_repo_overview`	Directorios de nivel superior, desglose de lenguajes, módulos principales
`search_code`	Resultados clasificados de archivos/símbolos para una consulta
`get_file_summary`	Tabla de símbolos, importaciones, exportaciones para un archivo
`get_dependencies`	Qué importa un archivo y quién lo importa
`get_relevant_context`	Paquete de contexto con presupuesto de tokens para una consulta, más una estimación de `token_savings` frente a un escaneo ingenuo del repositorio completo

Evaluar si esto realmente ayuda

eval/ contiene un benchmark pequeño y honesto contra el propio código de este repositorio (sin llamadas a LLM, sin números inventados): 10 consultas escritas a mano con archivos de verdad conocida, comparando nuestra herramienta indexada contra una línea base ingenua (listar el árbol, grep, leer archivos completos coincidentes).

.venv/bin/python eval/run_eval.py

Informa hit@1/hit@3 (¿apunta el resultado principal al archivo correcto?), reducción promedio de tokens y latencia. Esto solo mide la mecánica de recuperación/tokens — no mide si una respuesta real de Copilot es realmente mejor, ya que eso requiere llamadas a modelos en vivo.

Limitaciones actuales (MVP)

El análisis JS/TS se basa en regex (no en un AST completo), por lo que una sintaxis inusual puede pasarse por alto. El análisis de Python utiliza el módulo estándar ast y es exacto.
La búsqueda es solo léxica/de símbolos (con filtrado de palabras vacías y puntuación acumulada de múltiples señales); aún no hay embeddings/búsqueda semántica.
El observador de archivos aplica cambios por archivo pero no vuelve a recorrer los cambios de .gitignore en sí mismos en tiempo de ejecución — si se edita .gitignore, ejecuta index_repository con refresh=true una vez para recoger las nuevas reglas.

Consideraciones de seguridad antes de un uso más amplio/de producción

Ya corregido:

Inyección de shell — la extensión de VS Code anteriormente interpolaba ajustes del espacio de trabajo en cadenas de comandos de shell; ahora usa execFile con matrices de argumentos (sin shell) y se niega a ejecutar "Setup Server" en espacios de trabajo no confiables.
Escape de enlaces simbólicos — el recorredor omite enlaces simbólicos que se resuelven fuera de la raíz del repositorio (bloquea un enlace simbólico plantado para exponer archivos como /etc/passwd).
Fuga de secretos — los nombres de archivo que coinciden con patrones comunes de credenciales (.env, *.pem, id_rsa, credentials.json, etc., consulta SENSITIVE_FILENAME_PATTERNS en config.py) se omiten incluso si no están en gitignore, para que su contenido no pueda terminar en la salida de la herramienta.
Bloqueo por caché corrupta — un .mcp_intel_cache/index.json malformado/alterado ahora desencadena una reconstrucción limpia en lugar de bloquear el servidor al iniciarse.
ReDoS — el analizador regex JS/TS omite líneas individuales patológicamente largas (archivos minificados) para evitar DoS por retroceso catastrófico.
repo_root sin restricciones — establece MCP_INTEL_ALLOWED_ROOTS (una lista separada por : de rutas absolutas) para restringir qué directorios indexará el servidor; sin establecer por defecto para preservar el comportamiento flexible actual de un solo usuario.

Aún arquitectónico, no completamente resuelto — lee antes de implementar más allá de un solo usuario local:

No es seguro como servicio de red compartido/multiusuario. Esto está diseñado como un servidor stdio local de un proceso por usuario. Las cachés del índice/observador en memoria no tienen aislamiento por usuario ni autenticación. No expongas esto como un endpoint HTTP/SSE compartido sin añadir sandboxing y autenticación por llamador.
Las dependencias no están fijadas (solo >=) — fija versiones exactas o usa un archivo de bloqueo para instalaciones de producción reproducibles y verificadas (esto ya nos mordió una vez con un cambio de API incompatible de mcp 1.x → 2.0).
No hay pruebas de regresión automatizadas para este código base todavía — los cambios se verifican actualmente mediante el arnés manual eval/ y ejecuciones ad hoc, no una suite de pruebas controlada por CI.

Available Tools

6 tools

get_dependenciesA

Return what a file imports and which files import it (reverse deps), to understand blast radius before changing a file.

Name	Required	Description	Default
`path`	Yes
`repo_root`	No

TDQS

A3.5/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the burden. It states the operation is read-only ('return what a file imports...'), which is helpful, but it doesn't disclose whether the tool scans the whole repo, handles dynamic imports, or has performance implications. Given the lack of annotations, a 3 is appropriate.

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?

Two sentences with no fluff. The main purpose is stated first, and the use case is added compactly. Every word earns its place.

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?

For a tool with 2 params and no output schema, the description covers the primary use case, but it lacks details on repo_root semantics, return format, and any edge cases (like missing files). The complexity is moderate, so this is incomplete but not severely.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 0%, and the description does not explain either parameter. The path parameter is obvious from the tool name, but repo_root is undocumented. With no parameter info, a lower score might be expected, but the description's mention of 'file' implies path usage; still, the gap is significant.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool returns a file's imports and reverse dependencies, with a specific purpose (understanding blast radius). It distinguishes itself from siblings like search_code and get_file_summary by focusing on dependency relationships.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description implies usage ('to understand blast radius before changing a file') but does not explicitly state when not to use it or mention alternatives like get_relevant_context for broader context. The context signals show siblings, but no explicit routing guidance is given.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_file_summaryB

Return the symbol table (functions/classes/methods with signatures and docstrings), imports, and exports for a single indexed file — without returning the entire file body.

Name	Required	Description	Default
`path`	Yes
`repo_root`	No

TDQS

B3.4/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full behavioral burden. It discloses the main output content (symbol table, imports, exports) and a notable exclusion (not returning file body). However, it does not mention prerequisites like requiring the file to already be indexed or what happens when the file is missing.

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 a single well-structured sentence with no filler. It front-loads the core action and contents, then adds the clarifying exclusion. Every part contributes to understanding the tool.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given no output schema, no annotations, and zero parameter documentation, the description alone is not enough for an agent to confidently invoke this tool. It explains what is returned but not parameter semantics, indexing prerequisites, or error behavior, leaving important gaps.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema has 0% description coverage and the description does not explain 'path' or 'repo_root' semantics. While 'single indexed file' loosely ties 'path' to a file location, 'repo_root' remains unexplained, and no format, relative/absolute path guidance, or repository boundary details are provided.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description uses a specific verb ('Return') and names a concrete resource ('symbol table... imports, and exports') for a single indexed file. It clearly conveys what the tool does and even specifies what it doesn't do ('without returning the entire file body'), though it doesn't explicitly distinguish itself from siblings like get_relevant_context or get_repo_overview.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The phrase 'for a single indexed file' implies the tool is for per-file summaries, but there is no explicit guidance on when to prefer this over search_code, get_dependencies, or get_relevant_context. No exclusions or alternatives are mentioned, so usage context is only implied.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_relevant_contextA

Build a focused, token-budgeted context package (symbol tables + small code excerpts) for the files most relevant to a query. This is the main entry point for "help me find/fix X" style requests — use it instead of reading whole files or globbing the repo.

Name	Required	Description	Default
`query`	Yes
`repo_root`	No
`token_budget`	No

TDQS

A3.5/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description risks the full burden of behavioral disclosure. It communicates that the tool returns a synthesized context package rather than raw files, and that it is token-limited, but it does not state whether it is strictly read-only, whether it depends on an existing repository index, or what the actual response format looks like.

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 concise and well-structured. The first sentence states the core action and output, and the second provides routing guidance. There is no filler or repetition.

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 enough to orient an agent to the main purpose and broad usage, but it leaves gaps around the return structure, required prior state (e.g., indexation), and parameter behaviors. Because there is no output schema, more detail about the package shape would help.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 0%, and the description does not explain the three parameters. 'query' and 'token-budgeted' are lightly implied, but repo_root is entirely undocumented, meaning an agent cannot infer the expected input format or behavior for that parameter without external knowledge.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool builds a focused, token-budgeted context package of symbol tables and code excerpts for relevant files. It distinguishes itself as the 'main entry point' for find/fix requests, which differentiates it from lower-level or broader repository tools, though it does not explicitly name all sibling tools.

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?

It explicitly says to use this tool for 'help me find/fix X' requests and to use it instead of reading whole files or globbing the repository. This provides a clear when and when-not, but it does not name specific alternative tools such as search_code or get_file_summary.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_repo_overviewA

Return a high-level architecture summary: top-level directories, language breakdown, and the files with the most symbols (likely core modules). Use this to orient before drilling into specific files.

Name	Required	Description	Default
`repo_root`	No

TDQS

A3.8/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden. It discloses the read-only nature by saying 'Return' and describes the output shape, but it does not mention potential prerequisites, performance implications, or what happens when repo_root is null. This is adequate but incomplete.

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?

Two tight sentences with no filler. The first sentence front-loads the core behavior and output content; the second gives the usage context. Every word earns its place.

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?

For a simple read-only tool with one optional parameter and no output schema, the description covers the main purpose and output well. However, the undefined parameter behavior and the relationship to index_repository leave enough ambiguity that an agent may not know how to invoke it correctly on first use.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema has 0% description coverage for repo_root, and the tool description does not mention the parameter at all. The agent is left to infer that repo_root is the repository path and that null may mean the current directory, which the description does not clarify.

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 states exactly what the tool returns: a high-level architecture summary with top-level directories, language breakdown, and files with the most symbols. This clearly distinguishes it from sibling file-level tools like get_file_summary and search_code.

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?

It explicitly says to use this to 'orient before drilling into specific files,' which gives a clear when-to-use context. It does not name alternatives or state when not to use it, but the orientation purpose is specific enough.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

index_repositoryA

Index (or re-index) a repository: extracts files, symbols, imports and a reverse dependency graph. Call this first, and again with refresh=true after significant file changes. Uses an on-disk cache for fast re-runs.

Name	Required	Description	Default
`refresh`	No
`repo_root`	No

TDQS

A4.2/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavior disclosure, and it does reasonably well: it declares persistence behavior ('on-disk cache for fast re-runs'), signals re-indexing with refresh=true, and describes the data that is extracted. It stops at details like what the tool returns or whether it is non-destructive, but the core behavioral traits are covered.

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?

Three sentences, no filler: the first defines what it does, the second gives the call-order rule applied, and the third explains caching. Important behavioral information is front-loaded, and every clause adds value.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given a simple two-parameter schema with no output schema and no annotations, the description is close to complete: it asserts when to call it for the first time, when to re-run with refresh, what the internal behavior is (dependency graph building and caching), and what gets updated. It could specify what happens after indexing or what success looks like, but it is complete enough for this kind of setup tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 0%, so the description must compensate. It does explain refresh past the schema by tying it to re-running after file changes. However, repo_root is never mentioned or defined; an agent can infer it is the repository path, but the uncertainty is left to guesswork. Since at least one parameter is semantically enriched and the other is left implicit, this is adequate but not strong.

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?

States a specific action and resource: 'index (or re-index) a repository' and enumerates what it extracts — files, symbols, imports, and a reverse dependency graph. It reads as an explicit foundation step, and the siblings (get_repo_overview, search_code, get_dependencies) are clearly different read/query tools rather than indexing tools.

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 says 'Call this first, and again with refresh=true after significant file changes.' This is direct when-to-use guidance, making the invocation very and rerun conditions clear. It does not explicitly contrast with the sibling tools, but the relationship is strongly implied: this is the preliminary indexing step before querying with the get_* siblings.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

search_codeA

Search the indexed repository for files/symbols relevant to a natural language or keyword query (e.g. "authentication middleware", "parseConfig"). Returns ranked file hits with the matching symbol/reason — much cheaper than scanning the whole repo.

Name	Required	Description	Default
`query`	Yes
`top_k`	No
`repo_root`	No

TDQS

A3.5/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the burden of behavioral disclosure. It does reveal the main output behavior—ranked file hits with matching symbol/reason—and the cost advantage. However, it does not mention whether the repository must already be indexed, how repo_root is scoped, or any failure/edge behavior.

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 compact and front-loaded: it states the action, example usage, return behavior, and cost benefit in two efficient sentences. There is no fluff or redundant restatement of the tool name.

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?

This is a fairly simple tool and the description covers the core purpose and return shape adequately. Still, it leaves important invocation details undocumented, especially the semantics of top_k and repo_root, plus the implied prerequisite that the repository must be indexed first.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description adds useful meaning for the query parameter with natural-language/keyword guidance and examples, but top_k and repo_root are not explained at all. Since schema description coverage is 0%, the description was expected to compensate for all parameters but only covers one meaningfully.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool searches the indexed repository for files/symbols matching a natural language or keyword query and returns ranked file hits with matching symbols/reasons. It is specific about the resource and output, though it does not explicitly distinguish it from siblings like get_relevant_context.

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 when to use the tool: to find relevant files/symbols via semantic or keyword search. It also notes the tool is much cheaper than scanning the whole repo, which is useful guidance, though no explicit when-not-to-use or alternative routing is provided.

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.

6 tool updatesv0.1.0Aug 23, 2026
- First observedget_dependencies
- First observedget_file_summary
- First observedget_relevant_context
- First observedget_repo_overview
- First observedindex_repository
- First observedsearch_code

Deploy Server

Claude Code, inside the Godot 4 editorFlockbay puts an AI agent in the real Godot 4 editor: it writes GDScript, runs your game and fixes what breaks. Free, or bring your own Claude Code or Codex.flockbay.comAd

TDQS

A3.9/5.0

Scored 2026-08-23 00:25 across 6 tools

Disambiguation4/5

Each tool has a largely distinct purpose (index, overview, search, file details, dependencies, context), and the descriptive text flags get_relevant_context as the main entry point. However, search_code and get_relevant_context could both serve as the starting point for a 'find X' request, creating minor selection ambiguity.

Naming Consistency5/5

All tools follow a consistent verb_noun snake_case pattern (index_, get_, search_), making the API predictable and scannable. The only minor nitpick is abbreviating 'repository' as 'repo' in get_repo_overview, but this does not affect the overall pattern.

Tool Count5/5

Six tools sit at the sweet spot for a code-intelligence server, each earning its place in the index-query workflow. The set covers setup, orientation, discovery, drill-down, and impact analysis without bloat.

Completeness4/5

The core indexing-to-context pipeline is well covered, including re-indexing via refresh and reverse-dependency analysis for safe edits. Minor gaps exist (no full-file body reader, no way to list indexed repositories or explicitly purge the cache), but these seem like intentional scope decisions for a token-budget-focused tool.

Maintenance

ActivityStale

ResponsivenessNo recent community issues

Resources

Related MCP Connectors

Maguyva
ai.maguyva
Code intelligence for coding agents: semantic, AST, graph, and full-text search. 279+ languages.
Code Analysis Coding Agents
23
https://maguyva.tools/mcp
searchcodeA
com.searchcode
Code intelligence for LLMs. Analyze, search, and retrieve code from any public git repository.
Search Code Analysis Version Control
72
https://api.searchcode.com/v1/mcp
Graphify
io.github.Graphify-Labs
Search indexed code, trace dependencies, assess change impact, and recall repository memory.
Code Analysis Knowledge & Memory
1
https://api.graphify.com/mcp
Mapnostics
io.github.ffteema
Codebase intelligence for AI agents — dead code, blast radius, ownership.
Code Analysis Developer Tools
1
https://www.mapnostics.com/api/mcp

Related MCP Servers

Semantic Code Search MCPofficial
Code Analysis Developer Tools
elastic
A
license
Not graded
quality
D
maintenance
Provides AI coding agents with structured access to indexed codebases via semantic search, symbol analysis, and file reading tools.
Updated a day ago (2026-11-04 23:56 UTC)
13
Server Side Public , v 1
lens
Code Analysis Coding Agents
tools-for-agents
A
license
Not graded
quality
C
maintenance
Provides token-efficient code retrieval for coding agents by indexing repositories and enabling ranked snippet search, symbol outlines, and surgical line reads.
Updated a month ago (2026-09-27 08:59 UTC)
7 npm
MIT
codebase-rag
Code Analysis Developer Tools
Ar-maan05
A
license
Not graded
quality
D
maintenance
Enables LLM agents to efficiently understand and navigate a codebase by providing semantic search over symbols and a reference graph, replacing expensive grep/glob calls with structured tools like definition lookup, caller/callee queries, and change-impact analysis.
Updated 4 months ago (2026-07-21 19:03 UTC)
3
MIT
codegraph-mcp
Code Analysis Developer Tools RAG Systems
chainchen1103
F
license
Not graded
quality
A
maintenance
Provides efficient code navigation and graph-based analysis for AI agents, enabling symbol resolution, callers, implementations, and type schemas with minimal token usage.
Updated 2 months ago (2026-09-21 14:27 UTC)
-

Contexto de Inteligencia MCP

Por qué

Cómo funciona

Estructura del repositorio

Inicio rápido (Nuevos usuarios)

Ejecutar el servidor MCP de forma independiente

Instalación sin clonar este repositorio

Registrar con un cliente MCP (por ejemplo, VS Code)

Extensión de VS Code

Herramientas MCP disponibles

Evaluar si esto realmente ayuda

Limitaciones actuales (MVP)

Consideraciones de seguridad antes de un uso más amplio/de producción

Available Tools

Tool Schema Changelog

Maintenance

Resources

Related MCP Connectors

Related MCP Servers

Semantic Code Search MCPofficial