Complexity Source MCP
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| HOST | No | Host for HTTP development transport (default: 127.0.0.1). | |
| PORT | No | Port for HTTP development transport (default: 8790). | |
| MONGODB_URI | No | MongoDB connection string for fantasy catalog storage. | |
| GITHUB_TOKEN | No | GitHub token for reading private repositories (fine-grained, read-only). | |
| SOURCE_ROOTS | Yes | Absolute path(s) to allowed source directories, separated by OS path separator (e.g., ':' on macOS/Linux). | |
| MONGODB_DATABASE | No | MongoDB database name for fantasy catalog. | complexity_fantasy |
| SOURCE_MCP_TOKEN | No | Bearer token for accessing the MCP server in private Spaces. | |
| SOURCE_ALLOWED_HOSTS | No | Comma-separated list of allowed web hosts for web source tools. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| read_web_sourceA | Reads a public HTTP(S) page as text and returns its URL, retrieval time, SHA-256 fingerprint, and a bounded excerpt. Private network addresses are blocked. |
| read_github_sourceB | Reads one file from a GitHub repository using the GitHub contents API. GITHUB_TOKEN is optional for public repositories and required for private ones. |
| read_local_sourceA | Reads one UTF-8 text file located inside SOURCE_ROOTS and returns canonical provenance. Symbolic-link escapes are rejected. |
| search_local_sourcesA | Searches filenames and bounded UTF-8 files inside SOURCE_ROOTS. Build outputs, dependency folders, and Git metadata are skipped. |
| read_image_sourceA | Returns a PNG, JPEG, GIF, or WebP image from an allowed local path or public URL, together with verifiable provenance. The connected AI must itself support image content to interpret it. |
| list_source_policyA | Lists the configured read-only roots, host allowlist, source limits, and provenance guarantees. |
| search_fantasy_catalogA | Searches canonical fantasy entity cards stored in the managed catalog. Results include facts, relations, and provenance and never modify the catalog. |
| get_fantasy_entityB | Reads one canonical fantasy entity card and its directly related cards by stable key. |
| trace_fantasy_relationsA | Traverses a bounded relation graph around one canonical fantasy entity card for multi-entity reasoning. |
| fantasy_catalog_statusA | Returns only aggregate readiness and entity counts for the managed fantasy catalog. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| source-policy | Read-only source access, provenance fields, limits, and configured roots. |
TDQS
Scored across 10 tools
Every tool serves a clearly distinct purpose. Source tools are separated by source type (web, GitHub, local, image, policy), and fantasy tools are distinct in their operations (search, get, trace, status). No overlap.
All tool names follow a consistent verb_noun pattern (read_*, search_*, list_*, get_*, trace_*, *status). The naming is predictable and clear.
Ten tools is an appropriate number for a server covering two well-defined domains (source reading and fantasy catalog). Each tool earns its place without being excessive or insufficient.
The tool surface is complete for its intended read-only scope. Source reading covers all major source types with policy listing, and the fantasy catalog covers search, retrieval, relation tracing, and status. No obvious gaps given the stated purpose.