tibet-phantom-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PHANTOM_URL | No | Phantom server URL | http://localhost:8000 |
| PHANTOM_TIMEOUT | No | HTTP timeout in seconds | 30 |
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": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| phantom_statusA | Check Phantom server status — uptime, sessions, backends. |
| phantom_sessionsA | List all Phantom sessions (sealed and active) with IDs, descriptions, and timestamps. |
| phantom_backendsA | List available compute backends — local GPU, Vertex AI, Ollama, with models and latency. |
| phantom_sealA | Seal a session for cross-device resume. Creates a sealed Phantom session with full context. Resume on any device: curl -s $PHANTOM_URL/phantom/resume | sh Args: task: What you're working on description: Session description backend: Compute backend (p520-local, vertex-gemini, vertex-claude) model: AI model to use conversation: Message history [{"role": "user", "content": "..."}] todos: Todo items [{"content": "task", "status": "pending"}] files: Files to carry over {"name": "content"} packages: Pip packages needed target_identity: JIS identity for the session ttl_minutes: Session time-to-live Returns: Session ID and TIBET seal token |
| phantom_forkA | Fork into a session — inject an intervention (multi-AI handoff). Any actor can inject a signed message into a sealed session. The fork becomes part of the conversation with TIBET provenance. Use cases: correct AI mistakes, add context, human approval, cross-AI collaboration. Args: session_id: Target phantom session ID intervention: Message to inject actor: JIS identity (e.g., "jis:agent:root_ai", "jis:human:jasper") intent: Why (e.g., "correct_misconception", "add_context", "approve") Returns: Fork ID and TIBET token with ERIN/ERAAN/EROMHEEN/ERACHTER provenance |
| phantom_auditA | Full forensic audit of a session (Open Blackbox). Chronological events: who did what, when, why. All actors, backends, forks. Complete transparency into what happened inside an AI session. Args: session_id: Session to audit |
| phantom_fork_historyA | History of all forks/interventions in a session. Every external intervention chronologically with TIBET provenance per fork. Args: session_id: Session to get fork history for |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 7 tools
Each tool has a clearly distinct purpose: audit, list backends, fork, fork history, seal, list sessions, and status. No two tools overlap in functionality.
All tool names follow the consistent pattern 'phantom_<noun>' or 'phantom_<noun_noun>' in snake_case, making them predictable and easy to distinguish.
7 tools is well-scoped for a session management server covering creation, listing, forking, auditing, and status checks. Not too many or too few.
Core session lifecycle is covered: create (seal), list, fork, audit, and status. Missing a dedicated 'get_session' tool, but audit provides full details. Minor gap.