SagaSmith Narrative MCP
Enables long-form tabletop narrative play with durable campaigns, profiles, conflict phases, and branching storylines. Provides session-scoped MCP tools for lobby and play phases, including administrator controls for access, snapshots, and branches.
README
SagaSmith Narrative MCP
An authoritative, system-neutral MCP for long-form tabletop narrative play. It
depends on sagasmith-core for durable campaigns, transactions, revisions,
documents, continuity ledgers, snapshots, and branches.
The base runtime has two phases: lobby and play. A profile may opt into an
authoritative conflict phase. Native MCP tools are session-scoped and dynamic;
hosts must process tools/list_changed notifications.
The server never guesses rules from prose. A campaign binds an immutable profile version and checksum. Profiles may use Level 0 (explicit Agent/human rulings) or the limited pure Level 1 mechanics in this package. More complex rules require a separate system provider.
Lobby administrators can inspect campaign-scoped profile/Pack drafts through
narrative_query. Access changes support grant and revoke operations with
last-owner protection; controlled actors can be updated through actor_change.
Recovery queries and snapshot/branch mutations are administrator-only.
See Architecture and authority and
Profile and Pack lifecycle for the durable product
boundary. Three self-authored regression campaigns live in fixtures/ash-harbor,
fixtures/moss-road-seasons, and fixtures/echo-manor-voices. The third is a
15-session, five-NPC character-performance campaign with declared goals,
private motives, red lines, false beliefs, relationship arcs, voice markers,
isolated dialogue, and a recovered alternate ending.
Development
python -m pip install -e ".[dev]"
pytest
ruff check .
Run the original campaign fixtures concurrently through real stdio MCP sessions:
python scripts/regression_parallel_campaigns.py --output .runs/parallel
The runner opens real stdio MCP sessions, uses a separate session identity for each principal, executes every declared route step, follows a focused alternate branch, and emits machine-readable per-campaign timelines and a combined summary. A non-zero exit means the run is not accepted.
The Agent Host integration uses the MCP and Agent repositories' own environments: the
test process runs with ../SagaSmith-agent/.venv/Scripts/python.exe, while the
spawned MCP server runs with this repository's .venv/Scripts/python.exe:
../SagaSmith-agent/.venv/Scripts/python.exe -m pytest -q tests/test_agent_host_integration.py
Run locally with sagasmith-narrative-mcp. Its independent default home is
~/.sagasmith/narrative-mcp. Set SAGASMITH_NARRATIVE_MCP_HOME to relocate it,
SAGASMITH_NARRATIVE_MCP_DATABASE_URL to use an explicit database, and
SAGASMITH_NARRATIVE_MCP_BOUND_PRINCIPAL_ID when the transport authenticates
one principal.
Without SAGASMITH_NARRATIVE_MCP_BOUND_PRINCIPAL_ID, stdio is a trusted
single-user local mode; model-supplied principal fields are not multiplayer
authentication. A multiplayer deployment must bind one authenticated principal
per MCP process through a trusted transport. Shared-principal HTTP exposure is
not currently supported and the server must not be published directly to a
network.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.