TraceLattice

An MCP server that gives AI agents structured sequential thinking. Thoughts live in a DAG, reasoning strategies are pluggable, and confidence scores get calibrated against actual outcomes.

Features

• 11 thought types: regular, hypothesis, verification, critique, synthesis, meta, tool_call, tool_observation, assumption, decomposition, backtrack
• DAG-based thought graph with 8 edge kinds (sequence, branch, merge, verifies, critiques, derives_from, tool_invocation, revises) and topological traversal
• Pluggable reasoning strategies. Sequential by default, or Tree-of-Thought with BFS/beam search and plateau detection
• Tool interleave: suspend a thinking chain, run a tool call, then resume where you left off
• Confidence calibration using Beta(2,2) priors, Brier score, and Expected Calibration Error (ECE)
• Branch compression: cold branches get rolled into summaries automatically, with a sliding-window dehydration policy
• Outcome recording for tool_call/tool_observation results with metadata
• Tool and skill recommendations with confidence scores, rationales, and automatic discovery
• Per-session isolation with TTL eviction and LRU caching
• Three transports: stdio (default), SSE (legacy), Streamable HTTP (production)
• Strict TypeScript, Valibot validation, 1913 tests, 18-service DI container

Install

Requires Node.js v22+.

npm install -g tracelattice

Configure your MCP client

The default transport is stdio. Add the server to your client:

Claude Code

User-scoped (~/.claude.json) or project-scoped (.mcp.json in project root):

{
  "mcpServers": {
    "tracelattice": {
      "command": "tracelattice"
    }
  }
}

Or via CLI:

claude mcp add tracelattice -- tracelattice

Codex CLI

User-scoped (~/.codex/config.toml) or project-scoped (.codex/config.toml):

[mcp_servers.tracelattice]
command = "tracelattice"

Or via CLI:

codex mcp add tracelattice -- tracelattice

OpenCode

Global (~/.config/opencode/opencode.json) or project-scoped (.opencode.json):

{
  "mcpServers": {
    "tracelattice": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "tracelattice"
      ],
      "enabled": true,
      "environment": {
        "MAX_HISTORY_SIZE": "10000"
      }
    }
}

Configuration

Server

Variable	Default	Description
MAX_HISTORY_SIZE	1000	Maximum thoughts to keep in history
MAX_BRANCHES	50	Maximum number of branches
MAX_BRANCH_SIZE	100	Maximum size of each branch
LOG_LEVEL	info	Log level: debug, info, warn, error
PRETTY_LOG	true	Enable pretty log output

Feature flags

All flags default to false. Set to true or 1 to enable.

Variable	Description
TRACELATTICE_FEATURES_DAG_EDGES	Enable DAG edges for thought relationships
TRACELATTICE_FEATURES_CALIBRATION	Enable confidence calibration with Beta(2,2) priors
TRACELATTICE_FEATURES_COMPRESSION	Enable branch compression for cold branches
TRACELATTICE_FEATURES_TOOL_INTERLEAVE	Enable suspend/resume for tool calls
TRACELATTICE_FEATURES_NEW_THOUGHT_TYPES	Enable tool_call, tool_observation, assumption, decomposition, backtrack
TRACELATTICE_FEATURES_OUTCOME_RECORDING	Enable outcome recording for tool results
TRACELATTICE_FEATURES_REASONING_STRATEGY	Strategy: sequential (default) or tot

Transport

Variable	Default	Description
TRANSPORT_TYPE	stdio	Transport: stdio, sse, or streamable-http
STREAMABLE_HTTP_PORT	3000	Port for Streamable HTTP server
STREAMABLE_HTTP_HOST	localhost	Host for Streamable HTTP server
STREAMABLE_HTTP_STATEFUL	true	Enable stateful session tracking
SSE_PORT	3000	Port for SSE server
SSE_HOST	localhost	Host for SSE server
SSE_ENABLE_POOL	true	Enable connection pool for session isolation
SSE_MAX_SESSIONS	100	Maximum concurrent SSE sessions
SSE_SESSION_TIMEOUT	300000	SSE session timeout (ms)
CORS_ORIGIN	*	CORS origin
ENABLE_CORS	true	Enable CORS preflight
ALLOWED_HOSTS	(all)	Comma-separated allowed hosts

Skill discovery

Variable	Default	Description
SKILL_DIRS	(none)	Colon-separated skill directories
DISCOVERY_CACHE_TTL	300000	Discovery cache TTL (ms)
DISCOVERY_CACHE_MAX_SIZE	100	Discovery cache max entries

Transports

Set TRANSPORT_TYPE to pick one:

Transport	When to use	Command
stdio (default)	Local MCP clients	tracelattice
sse (legacy)	Multi-user setups, backwards compatibility	TRANSPORT_TYPE=sse tracelattice
streamable-http	Production deployments	TRANSPORT_TYPE=streamable-http tracelattice

Development

npm install
npm run dev          # MCP inspector
npm test             # 1913 tests (vitest)
npm run test:coverage # with coverage report
npm run type-check   # tsc --noEmit
npm run lint         # eslint
npm run build        # rslib + rsbuild

Architecture

src/
├── core/               # Domain logic
│   ├── graph/          # DAG edges: Edge, EdgeStore, GraphView
│   ├── evaluator/      # SignalComputer, Aggregator, PatternDetector, Calibrator
│   ├── compression/    # CompressionService, DehydrationPolicy, SummaryStore
│   ├── reasoning/      # OutcomeRecorder + strategies (Sequential, TreeOfThought)
│   └── tools/          # InMemorySuspensionStore (suspend/resume)
├── contracts/          # Shared interfaces (cross-module coupling point)
├── persistence/        # File, SQLite, Memory backends
├── transport/          # stdio, SSE, Streamable HTTP
├── di/                 # IoC container (18 services)
├── registry/           # Tool/Skill discovery with LRU cache
└── config/             # YAML + env var loading

License

MIT