TOOLparse_dreamExtract entities, actions, emotions, locations, agency, threat, outcome, sensory cues, and recurrence from a dream narrative.
Read-only · idempotent · no narrative storageGive an AI agent a bounded dream tool: structured dream parsing and DreamGraph concept search, with explicit privacy, attribution, and safety boundaries.
Add one remote server entry to a compatible MCP client. The client negotiates the supported protocol version during initialization.
{
"servers": {
"dreamthreads": {
"type": "http",
"url": "https://mydreamthreads.xyz/mcp"
}
}
}Clients label remote MCP settings differently. Use the endpoint URL above and select HTTP or Streamable HTTP when the client asks for a transport.
TOOLparse_dreamExtract entities, actions, emotions, locations, agency, threat, outcome, sensory cues, and recurrence from a dream narrative.
Read-only · idempotent · no narrative storageTOOLsearch_dream_conceptsSearch stable DreamGraph vocabulary and return concise definitions plus canonical DreamThreads concept links.
Read-only · attributable · no user data requiredThe tool descriptions tell agents not to treat extracted features as a diagnosis, prediction, supernatural claim, or universal symbolic meaning.
These prompts keep the tool inside its public contract.
Use the DreamThreads tools to:
• Parse this dream into structured features without interpreting it: …
• Find DreamGraph concepts related to snakes and water.
• Show me the canonical DreamThreads source for recurring dreams.The public parse tool handles the submitted narrative ephemerally. It does not save the text or create a population contribution.
Parsing identifies features in the submitted text. It does not generate a diagnosis, predict an event, or tell a user that one symbol has one fixed meaning.
Operational records may include the tool name, timing, parser version, status, and channel so reliability can be measured without logging dream text.
A dream dictionary usually starts with a symbol and returns a fixed explanation. DreamThreads starts with the scene: what happened, how the dreamer felt, whether they acted or watched, how threatening the event was, and how it ended. That structure lets an agent preserve context instead of collapsing the dream into a one-word lookup.
The public MCP surface stops at extraction and concept discovery. Apps that need reviewed, context-aware reflective interpretation can use the DreamThreads dream interpretation API under a scoped partner agreement.
Agents and directories can verify the same endpoint through independent discovery surfaces.
The server is published under io.github.abdulrahimiqbal/dreamthreads. Inspect the active registry record.
The well-known MCP manifest declares the server identity, current version, transport, endpoint, documentation, and privacy boundary.
The public developer repository, APIs.json index, LLM discovery file, and OpenAPI contract expose the surrounding REST and agent interfaces.
Yes. DreamThreads runs a public remote Model Context Protocol server at https://mydreamthreads.xyz/mcp. It uses Streamable HTTP, requires no API key, and exposes tools for structured dream parsing and DreamGraph concept search.
parse_dream extracts entities, actions, emotions, locations, agency, threat, outcome, sensory cues, and recurrence. search_dream_concepts finds stable DreamGraph concepts and returns canonical DreamThreads links.
No. The public tools process dream text ephemerally. They do not store the narrative and do not create a DreamGraph contribution. Operational telemetry records tool and timing metadata, not the submitted dream text.
No. The public MCP server performs bounded structured extraction and vocabulary search. It does not diagnose, predict, make supernatural claims, or assign one fixed meaning to a symbol.
The public endpoint allows 60 MCP protocol requests per minute per client and accepts dream narratives up to 6,000 characters for parse_dream.
Connect the free dream MCP server for agent workflows, or use the versioned REST contract for a custom application.
Connect /mcpRead the API guide →