​
FL Repos Intelligence Platform

FL Repos turns source repositories into a queryable map of code, ownership, architecture, security signals, dependencies, reports, and improvement opportunities. It began with ForwardLane’s own portfolio; it now supports a fully open-source local Community Edition and a multi-tenant hosted workspace for organization-scoped operations. The same analysis model powers both surfaces, but their boundaries differ. Local users operate on their own repositories and data. Hosted users work inside an organization-scoped workspace; its browser sessions, credentials, report artifacts, usage, and billing records are not a shared global knowledge graph. Read Product Model first if you need to choose between local and hosted workflows.

​
Choose your path

​
The Knowledge Graph

The graph is stored in FalkorDB for graph-backed deployments, with local data stores supporting Community Edition workflows. Its repository count and topology change as enrichment runs complete; use GET /api/stats for the current operations snapshot. The model includes repository, source-file, code, API, database, test, contributor, security, dependency, and configuration nodes. Every repository, source file, function, class, API endpoint, database model, test case, contributor, security pattern, dependency, and config file is a first-class node in the graph. Edges encode the relationships between them — which repo owns which file, which function calls which other function, which contributor committed to which repo, and so on.

​
Node Types at a Glance

Repository

Repositories with metadata such as language, tier, commit count, contributor count, and score.

Function

Named functions with complexity, argument lists, decorators, and docstrings.

Class

Python/JS classes with base classes, method counts, and line counts.

SourceFile

Every analyzed source file with language, line count, and entity counts.

APIEndpoint

Discovered HTTP endpoints with path, method, view name, and URL pattern.

DataModel

Django ORM models with field counts and meta options.

ModelField

Individual model fields with type, nullability, index, and FK targets.

TestCase

Test functions and classes with test type classification.

Contributor

Git authors with commit counts, repo counts, and activity windows.

SecurityPattern

Detected security patterns (hardcoded secrets, insecure calls) with severity.

Dependency

External packages imported by repos (PyPI, npm, etc.).

ConfigFile

Configuration files (Docker, CI/CD, environment, framework configs).

​
Deep Insights

The platform goes beyond simple repository metadata. Its deep-insight layer extracts fine-grained information from source code:

​
Functions

Every named function extracted from Python and JavaScript source files, annotated with:
  • Complexity score — cyclomatic complexity estimate
  • Argument list — parameter names
  • Decorators — e.g., @login_required, @cached_property
  • Docstring — the first docstring found in the function body
  • Line count — size of the function

​
Classes

Every class definition with its inheritance chain, method count, and documentation.

​
Django Models

Django ORM model classes with their full field definitions — field types, nullability, indexing, foreign key targets — enabling cross-repo schema analysis.

​
API Endpoints

HTTP endpoints discovered via Django URL conf parsing, giving you a complete API surface map across all services.

​
Contributors

Every Git author found in an analyzed repository, with commit counts, repo counts, and activity timestamps — enabling contributor network analysis.

​
Security Patterns

Static-analysis detections of security-relevant patterns: hardcoded credentials, use of eval(), SQL string formatting, and other high-severity issues. Each finding is tagged with a severity level.

​
Access Methods

​
REST API

The REST API is the primary programmatic interface. It runs at:
Protected operations routes use a Bearer token. Public exceptions are GET /, GET /health, and GET /v1/billing/plans. The API returns JSON and supports pagination, filtering, and sorting on most list endpoints.

​
CLI

The supported CLI is fl-repos. It runs local extraction, enrichment, graph, fleet, report, and tenant-cloud workflows; it is not a wrapper for every REST endpoint.

​
MCP Server

The MCP (Model Context Protocol) server exposes the graph as a set of tools that AI agents can call. It supports two transports:
  • stdio — for Claude Desktop integration
  • SSE — for network clients at https://fl-mcp-production.up.railway.app
When Claude Desktop is configured with the MCP server, you can ask questions like “Which repos have the most security vulnerabilities?” or “Show me all Django models related to billing” and Claude will call the appropriate graph tools to answer.

​
Repository Tiers

Repositories are classified into tiers based on a composite score (commit count, contributor count, recency, and other signals): Use the tier filter on /api/repos or the --tier flag on the CLI to narrow results by tier.