MCP Tools

The FL Repos MCP server exposes 15 tools split into read tools (always available) and write tools (require --write flag).

Read Tools

discover

Get the graph schema, platform statistics, node/edge type descriptions, and example Cypher queries. Call this first in any session to orient the AI agent. Parameters: None Returns: A structured document containing:

Node type list with descriptions and key properties
Edge type list with source/target information
Current graph statistics (node count, edge count, repo count)
5–10 example Cypher queries demonstrating common patterns

Example invocation (Claude Desktop):

What does the FL Repos graph contain?

Claude calls discover and summarizes the schema for you.

query

Execute a read-only Cypher query directly against the FalkorDB graph. This is the most powerful tool — any traversal, aggregation, or pattern match is possible. Parameters:

Parameter	Type	Required	Description
`cypher`	string	Yes	The Cypher query to execute. Must be a read query (MATCH/RETURN).
`limit`	integer	No	Override the LIMIT clause (max 500).

Returns: Query results as a list of row objects. Constraints: Only MATCH, RETURN, WHERE, WITH, ORDER BY, LIMIT, UNWIND, OPTIONAL MATCH are permitted. Mutation keywords are rejected. Example queries:

-- Top 10 repos by score
MATCH (r:Repository)
RETURN r.name, r.score, r.tier
ORDER BY r.score DESC
LIMIT 10

-- Functions with highest complexity
MATCH (f:Function)
WHERE f.complexity > 15
RETURN f.name, f.repo, f.complexity
ORDER BY f.complexity DESC
LIMIT 20

-- Cross-repo contributor activity
MATCH (c:Contributor)-[:COMMITTED_TO]->(r:Repository)
WITH c, collect(r.name) AS repos, count(r) AS repo_count
WHERE repo_count >= 5
RETURN c.name, repo_count, repos
ORDER BY repo_count DESC

search_repos

Search and filter the repository list. A higher-level alternative to query for common repo lookup patterns. Parameters:

Parameter	Type	Required	Description
`query`	string	No	Keyword search on name and description.
`language`	string	No	Filter by primary language.
`tier`	integer	No	Filter by tier (1, 2, or 3).
`min_commits`	integer	No	Minimum commit count.
`sort_by`	string	No	Sort field: `name`, `score`, `commits`.
`limit`	integer	No	Max results (default 20).

Returns: List of matching repository objects with name, language, tier, score, and commit count. Example:

Show me all Tier 1 Python repos sorted by score.

Claude calls search_repos(tier=1, language="python", sort_by="score", limit=20).

get_repo_detail

Get the full metadata and statistics for a single repository. Parameters:

Parameter	Type	Required	Description
`name`	string	Yes	Repository name.

Returns: Full repository object including function count, model count, endpoint count, test count, security issue count, language percentages, and all metadata fields. Example:

Tell me everything about the forwardlane-backend repository.

get_recommendations

Get tier-based repository recommendations. Returns the top repos from each tier, useful for onboarding new team members or getting a quick overview of the most important repos. Parameters:

Parameter	Type	Required	Description
`tier`	integer	No	Restrict recommendations to a specific tier.
`top_n`	integer	No	Number of repos to return per tier (default 5).

Returns: Ranked list of repos with scores and brief descriptions.

list_functions

List functions from the deep-insight layer with optional filters. Parameters:

Parameter	Type	Required	Description
`repo`	string	No	Filter to a specific repository.
`name`	string	No	Filter by function name substring.
`min_complexity`	integer	No	Minimum cyclomatic complexity.
`limit`	integer	No	Max results (default 50).

Returns: List of function objects with name, repo, file path, args, decorators, docstring, line count, and complexity.

list_models

List Django ORM models with optional filters. Parameters:

Parameter	Type	Required	Description
`repo`	string	No	Filter to a specific repository.
`name`	string	No	Filter by model name substring.
`limit`	integer	No	Max results (default 50).

Returns: List of model objects with name, repo, file path, field count, has_meta flag, and docstring.

list_contributors

List contributors with activity metrics. Parameters:

Parameter	Type	Required	Description
`min_commits`	integer	No	Minimum total commits.
`min_repos`	integer	No	Minimum repos contributed to.
`limit`	integer	No	Max results (default 50).

Returns: List of contributor objects with name, email, total commits, repos count, and activity window.

list_security

List detected security pattern findings. Parameters:

Parameter	Type	Required	Description
`repo`	string	No	Filter to a specific repository.
`severity`	string	No	Filter by severity: `high`, `medium`, `low`.
`limit`	integer	No	Max results (default 50).

Returns: List of security pattern objects with type, repo, file path, severity, and description.

search_memory

Search the Graphiti or Zep memory backend for past episodes, facts, or conversations related to a query. Parameters:

Parameter	Type	Required	Description
`query`	string	Yes	Natural language search query.
`limit`	integer	No	Max results (default 10).

Returns: List of matching memory episodes with content, timestamp, and relevance score.

Memory search requires Graphiti or Zep to be configured. If neither backend is available, this tool returns an empty result with a configuration hint.

Write Tools

Write tools are only available when the MCP server is started with the --write flag. They permanently modify the graph. Use with caution.

add_entity

Add a new node to the graph. Parameters:

Parameter	Type	Required	Description
`label`	string	Yes	Node type label (e.g. `Repository`, `Function`).
`properties`	object	Yes	Key-value properties for the node.

Returns: The created node with its assigned graph ID.

update_entity

Update properties on an existing graph node. Parameters:

Parameter	Type	Required	Description
`node_id`	string	Yes	The node identifier to update.
`properties`	object	Yes	Properties to set or update.

Returns: The updated node.

add_relationship

Add an edge between two existing nodes. Parameters:

Parameter	Type	Required	Description
`source_id`	string	Yes	Source node identifier.
`target_id`	string	Yes	Target node identifier.
`relationship`	string	Yes	Edge type label (e.g. `DEPENDS_ON`, `HAS_FUNCTION`).
`properties`	object	No	Optional edge properties.

Returns: The created edge.

delete_entity

Delete a node and all its associated edges from the graph. Parameters:

Parameter	Type	Required	Description
`node_id`	string	Yes	The node identifier to delete.

Returns: Confirmation of deletion with count of edges removed.

Deletion is irreversible without a graph backup. Deleting a node also removes all edges connected to it. Only use this for truly orphaned or erroneous nodes.

add_memory

Add an episode to the configured Graphiti or Zep memory backend. Parameters:

Parameter	Type	Required	Description
`content`	string	Yes	The episode content (fact, conversation turn, observation).
`source`	string	No	Source label for the episode (e.g. `claude`, `user`, `system`).
`metadata`	object	No	Optional key-value metadata.

Returns: The stored episode with its assigned ID and timestamp. See Memory Backends for how Graphiti and Zep store and retrieve episodes.

Get Started

Products

API Reference

CLI

FL Web

Python SDK

MCP Server

Graph Schema

MCP Tools

MCP Tools

Read Tools

discover

query

search_repos

get_repo_detail

get_recommendations

list_functions

list_models

list_contributors

list_security

search_memory

Write Tools

add_entity

update_entity

add_relationship

delete_entity

add_memory

​MCP Tools

​Read Tools

​discover

​query

​search_repos

​get_repo_detail

​get_recommendations

​list_functions

​list_models

​list_contributors

​list_security

​search_memory

​Write Tools

​add_entity

​update_entity

​add_relationship

​delete_entity

​add_memory

MCP Tools

Read Tools

discover

query

search_repos

get_repo_detail

get_recommendations

list_functions

list_models

list_contributors

list_security

search_memory

Write Tools

add_entity

update_entity

add_relationship

delete_entity

add_memory