Liminality
Breaks a hard question or decision into checkable sub-questions, grounds each to a real tool.
What it can do
What data it sees
Do you need an account
An API key from the service settings is required
Breaks a hard question or decision into checkable sub-questions, grounds each to a real tool.
How to connect
How to connect Liminality to Claude
The server address to paste into Claude settings:
https://liminality.physea.ai/mcp
- Open Claude (claude.ai or the desktop app).
- Go to Settings → Connectors.
- Click Add custom connector.
- Paste the server address copied below into Remote MCP server URL and click Add.
- In a chat, click + → Connectors and switch the new connector on.
Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.
Authorization
After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.
How to connect Liminality to ChatGPT
The server address to paste into ChatGPT settings:
https://liminality.physea.ai/mcp
- Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
- Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
- Open Settings → Connectors and click Create.
- Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
- Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
- In a new chat open + → Apps/Connectors and enable the connector.
OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.
Authorization
On first use ChatGPT opens the service's sign-in window. Sign in and approve access.
How to connect Liminality to Cursor
The server address to paste into Cursor settings:
https://liminality.physea.ai/mcp
Fastest: click Open in Cursor below and confirm the prompt.
Manually:
- In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
- Paste the JSON copied below into
~/.cursor/mcp.json(per project:.cursor/mcp.jsonin the repo root). - Save the file. The server appears in the MCP list; authorize it there if asked.
Authorization
If the service needs sign-in, an authorize button appears next to the server in the MCP list.
Server tool list (27)
Raw names from tools/list. Only developers need these.
| solve | Decompose, ground, and answer a non-trivial task. Supports saved or per-call depth, safe trace, caller tools, and lean/full output. |
| get_capabilities | List solve modes and every public capability. |
| get_settings | Read this caller's saved solve and safety settings. |
| set_settings | Save caller-scoped settings from one dict (depth, optimization, trace, relay, governance); keys and values in get_capabilities. |
| get_result_detail | Fetch the full caller-owned result behind a lean solve. |
| export_result | Export a caller-owned result in a supported format. |
| compress_context | Compress long context into a task-focused brief; use before reading any source over 8,000 characters. |
| delegation | Create, resume, or report a bounded subagent packet (action='create'|'resume'|'report'). |
| handoff | Create, safely resume, or list caller-scoped cross-session checkpoints (action='create'|'resume'|'list'). Create accepts an optional subject and source_sessions; Liminality assigns the subject number. Resume by handoff_id — 'latest' resumes the newest unfinished resumable checkpoint, or asks which one when several were created in the last 24h; detail='full' returns the complete checkpoint. |
| ask_form | Create a short form for missing user-specific facts. |
| apply_form | Apply form answers to sharpen a task. |
| report_outcome | Report verified real-world outcome evidence for a result. |
| report_feedback | Report structured feedback on a result. |
| apply_grade | Apply a grade-form answer to a result. |
| optimize | Rank bounded one-factor variants for a route against your stated objective; candidates remain unsaved. |
| algorithms | Create, validate, release and run pure algorithms over versioned datasets; runs replay byte-identical with zero model calls. action='run' is one solve; other actions are free. |
| task | Read or version the contract behind a solve (action='get'|'list'|'versions'|'new_version'|'graph'|'create'|'state'); graph also returns what is still unfinished. |
| recover | Find a workaround for a blocked obligation without changing the task: open a case, read candidates with their boundary, choose, retry, verify against the original criteria (action='open'|'candidates'|'choose'|'retry'|'verify'|'add_candidate'|'get'|'list'). |
| runs | Run lifecycle with run and attempt identity (action='start'|'status'|'list'|'cancel'|'reconcile'|'attempt'|'execute'|'outcome'|'approve'|'grants'|'revoke'): cancel persists, an unknown effect is reconciled before any retry, simulate never executes, approve grants operation_kinds for hours and returns approval_id. |
| research | Run a multi-source research pass for a question. |
| composio_connect | Create a caller-scoped connection link for a toolkit. |
| register_asset | Register a caller-owned resource for future grounding. |
| set_priorities | Save caller-scoped tasks and importance signals. |
| taxis | Return this caller's deterministic ranked priority frame. |
| catalog | Read the capability catalog: action='search' returns every saved candidate for an objective with distinctiveness deltas; action='delta' compares a named entity pair. |
| setup | Context-minimality setup with no questions (auto); or profiles, status, verify. |
| manage | Compatibility router for retained-context lifecycle actions. |