astrovisor-mcp
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@astrovisor-mcpCalculate my current transits"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
AstroVisor MCP
AstroVisor MCP connects MCP-compatible AI clients to the complete AstroVisor API. It supports local stdio and production Streamable HTTP, discovers operations from the live OpenAPI document, and returns large calculation results through a compact, queryable response layer.
This repository is the protocol and server layer. For reusable personal workflows, private people profiles, interactive credential setup, and safe AI request construction, install the separate AstroVisor Skill.
AstroVisor calculations require a personal dashboard API key beginning withpk-. Anonymous calculation requests are not supported. Create an account and
get your API key at astrovisor.io, then provide it to
your MCP client through a secure environment variable or the AstroVisor Skill
private credential setup.
Configure the API key
For a direct MCP connection, make the key available to the client process:
export ASTROVISOR_API_KEY="pk-..."For the safer profile-aware setup, install AstroVisor Skill and use its hidden key prompt:
npx --yes --package=astrovisor-skill@1.0.1 -- \
astrovisor-skill install --target both
node "$HOME/.agents/skills/astrovisor/scripts/astrovisor-skill.mjs" \
config set-key
node "$HOME/.agents/skills/astrovisor/scripts/astrovisor-skill.mjs" \
doctor --jsonYou can also ask your agent:
Use $astrovisor. Help me securely configure my AstroVisor API key, keep it out of
chat and client configuration, then verify the MCP connection.The agent should guide you to the hidden terminal prompt rather than asking you to paste the full key into the conversation.
AstroVisor Skill (optional workflow layer)
↓
AstroVisor MCP (this repository)
live OpenAPI · authentication · API calls · result retrieval
↓
AstroVisor APIHighlights
Six compact MCP tools covering 456 current API operations.
Live OpenAPI discovery instead of hundreds of hardcoded tool definitions.
Stdio support for Claude Desktop, Claude Code, Codex, and other local clients.
Authenticated Streamable HTTP at
https://mcp.astrovisor.io/.Exact
operationId, path, query, and JSON body serialization.Common core and
birth_*request-profile normalization.Compact, full, and summary response views with token budgets.
Filtering, projection, sorting, pagination, and targeted path retrieval.
Temporary result storage isolated by caller API key.
Optional full mode with one generated tool per OpenAPI operation.
Capabilities include natal astrology, transits, progressions, solar returns, synastry, compatibility, Jyotish, BaZi, Human Design, Gene Keys, numerology, Tarot, Lenormand, astrocartography, horary, electional, and every other operation published by the live AstroVisor OpenAPI schema.
Related MCP server: OpenAPI MCP Server
Choose your integration
Need | Recommended setup |
Personal assistant with reusable profiles | |
Cloud or generic MCP client | Remote Streamable HTTP |
Claude Desktop or local stdio client |
|
Backend/operator deployment | Internal JSON-RPC adapter plus public gateway |
Direct Node.js dependency |
|
Requirements:
Node.js 20 or newer for local stdio;
an AstroVisor dashboard API key beginning with
pk-;an MCP client supporting stdio or Streamable HTTP.
Quick start
Remote Streamable HTTP
Endpoint:
https://mcp.astrovisor.io/Authenticate with either:
Authorization: Bearer pk-...or:
X-API-Key: pk-...The compatibility endpoint https://mcp.astrovisor.io/mcp is also available.
Health:
https://mcp.astrovisor.io/healthLocal stdio
Run without a global installation:
ASTROVISOR_API_KEY="pk-..." \
npx --yes --package=astrovisor-mcp@5.0.0 -- astrovisor-mcpOr install globally:
npm install --global astrovisor-mcp@5.0.0
ASTROVISOR_API_KEY="pk-..." astrovisor-mcpFor personal use, prefer the private launcher included in AstroVisor Skill. It avoids putting the key in MCP configuration or shell history.
Client configuration
Codex and ChatGPT desktop
Remote HTTP using an environment variable:
[mcp_servers.astrovisor]
url = "https://mcp.astrovisor.io/"
bearer_token_env_var = "ASTROVISOR_API_KEY"
startup_timeout_sec = 30
tool_timeout_sec = 120Codex CLI, the IDE extension, and ChatGPT desktop share Codex MCP configuration on the same host. Verify with:
codex mcp listIn the interactive client, use /mcp.
For profile-aware workflows and generated secret-free configuration, follow the AstroVisor Skill Codex setup.
Official documentation: Codex MCP.
Claude Code
Remote HTTP:
claude mcp add --transport http --scope user astrovisor \
https://mcp.astrovisor.io/ \
--header "Authorization: Bearer pk-..."Local stdio:
claude mcp add --transport stdio --scope user \
--env ASTROVISOR_API_KEY=pk-... \
astrovisor -- \
npx --yes --package=astrovisor-mcp@5.0.0 -- astrovisor-mcpVerify:
claude mcp get astrovisor
claude mcp listRemote HTTP is the preferred transport for a cloud-hosted server. For personal credentials, the AstroVisor Skill launcher keeps the key outside Claude configuration and shell history.
Official documentation: Claude Code MCP.
Claude Desktop
Direct stdio configuration:
{
"mcpServers": {
"astrovisor": {
"command": "npx",
"args": [
"--yes",
"--package=astrovisor-mcp@5.0.0",
"--",
"astrovisor-mcp"
],
"env": {
"ASTROVISOR_API_KEY": "pk-...",
"ASTROVISOR_URL": "https://astrovisor.io",
"ASTROVISOR_TOOL_MODE": "compact"
}
}
}
}Restart Claude Desktop after changing claude_desktop_config.json.
For a configuration with no embedded API key, use the AstroVisor Skill Claude Desktop setup.
Compact toolset
Compact mode is the default. It keeps tool definitions small while preserving access to every live API operation.
Tool | Purpose |
| Read global request, alias, and response conventions |
| Search operations by intent or keyword |
| List operations by tags, path, method, or page |
| Inspect one canonical operation and request schema |
| Execute any operation by |
| Retrieve a stored result or a narrow fragment |
The correct discovery chain is:
astrovisor_conventions
↓
astrovisor_openapi_search / astrovisor_openapi_list
↓
astrovisor_openapi_get
↓
astrovisor_request
↓
astrovisor_result_get (when a stored or narrower result is needed)Never guess an operation id or request body. astrovisor_openapi_get returns:
canonical
operationId;HTTP method and path;
path and query parameters;
requestBodySchema;aliases;
required body fields;
an LLM-oriented example body.
Request contract
Call astrovisor_request with this outer envelope:
{
"operationId": "<canonical operation id from astrovisor_openapi_get>",
"path": {},
"query": {},
"body": {},
"response": {
"view": "compact",
"tokenBudget": 12000,
"store": true
}
}Rules:
put URL-template variables only in
path;put query-string values only in
query;put the API JSON request only in
body;preserve booleans, numbers, arrays, and objects as JSON types;
do not wrap
bodyindata,payload, orrequestunless the live schema requires it;do not send empty strings as substitutes for required fields;
use the live schema when it differs from examples or cached knowledge.
The MCP normalizes common aliases between:
datetime, latitude, longitude, location, timezoneand:
birth_datetime, birth_latitude, birth_longitude,
birth_location, birth_timezoneNormalization is a compatibility feature, not a reason to skip live schema inspection.
For a strict personal and multi-person workflow, use the AstroVisor Skill request contract.
Large results and precision retrieval
Every compact response uses the astrovisor.serialized.v2 envelope and can include:
meta.query.totalBefore,totalMatched,offset,limit, andnextCursor;meta.availablePathsfor discoverable follow-up paths;meta.pathFoundfor path validation;meta.truncatedwhen additional retrieval is appropriate;summary.sourceandsummary.selected;a temporary
resultId;token-optimized
data.
Response controls:
Field | Purpose |
| Select a subtree before other processing |
| Project specific fields |
| Filter using field/operator expressions |
| Apply deterministic ordering |
| Continue cursor pagination |
| Offset pagination |
| Keep or omit paths |
| Bound large arrays |
| Bound serialized output |
| Keep the full response temporarily |
Example:
{
"operationId": "<operation id>",
"body": {
"<required field>": "<confirmed value>"
},
"response": {
"view": "compact",
"responsePath": "data.items",
"select": ["date", "strength", "theme"],
"where": {
"strength_gte": 0.75
},
"sort": ["-strength", "date"],
"responseLimit": 20,
"tokenBudget": 12000,
"store": true
}
}Retrieve the next fragment without recalculating:
{
"resultId": "<result id>",
"response": {
"view": "compact",
"responsePath": "data.items",
"cursor": "<next cursor>",
"select": ["date", "strength", "theme"],
"responseLimit": 20,
"tokenBudget": 12000
}
}Supported where suffixes:
_eq _ne _gt _gte _lt _lte
_in _nin
_contains _startswith _endswith
_exists _regexIf meta.pathFound is false, select a path from meta.availablePaths instead of
repeating the same request.
Full tool mode
Set:
export ASTROVISOR_TOOL_MODE=fullFull mode creates one MCP tool per OpenAPI operationId and accepts several legacy
operation aliases. It is useful for specialized clients but can consume too much
context in clients with strict tool-definition limits. Compact mode is recommended
for general use.
Environment variables
Stdio and internal adapter
Variable | Default | Purpose |
| required | Dashboard API key |
|
| API base URL |
|
| OpenAPI override |
|
|
|
|
| Default response view |
| implementation default | Serialized byte budget |
|
| Result cache TTL |
|
| Maximum cached results |
Public gateway
Variable | Default | Purpose |
|
| Gateway bind host |
|
| Gateway bind port |
|
| Internal adapter URL |
|
| Non-billable key validation endpoint |
|
| Validation cache time |
| required | Internal validation credential |
Self-hosting
The production topology is:
Remote MCP client
↓ TLS
Reverse proxy
↓ loopback
Public Streamable HTTP gateway
↓ loopback
Internal JSON-RPC adapter
↓
AstroVisor APIBuild and start the two Node services:
npm ci
npm run build
npm run start:jsonrpc
npm run start:publicOperational requirements:
keep the AstroVisor API, internal adapter, and public gateway on loopback;
terminate TLS at a reverse proxy;
require
MCP_INTERNAL_VALIDATION_TOKENfor key validation;ensure key validation does not record billable API usage;
apply process supervision, memory limits, and restart policies;
never log raw caller API keys or full private request bodies.
Development
git clone https://github.com/rokoss21/astrovisor-mcp.git
cd astrovisor-mcp
npm ci
npm run build
npm run test:unitProduction OpenAPI smoke test:
ASTROVISOR_URL=https://astrovisor.io npm testStdio end-to-end test:
ASTROVISOR_API_KEY=pk-... npm run test:e2e:stdioRemote end-to-end test:
MCP_URL=https://mcp.astrovisor.io/ \
ASTROVISOR_API_KEY=pk-... \
npm run test:e2e:remoteSecurity check:
npm audit --omit=devMigration from 4.3.x
AstroVisor Skill now has its own canonical repository and npm package:
GitHub: rokoss21/astrovisor-skill
npm:
astrovisor-skill
Install it directly:
npx --yes --package=astrovisor-skill@1.0.1 -- \
astrovisor-skill install --target bothVersion 5.0.0 removes the bundled astrovisor-skill binary. This avoids npm
binary-name collisions and gives each project a single release lifecycle. Replace
commands that install the skill through astrovisor-mcp with the standalone
package command above. The MCP tool and transport contracts are unchanged.
Related projects
AstroVisor Skill — profiles, onboarding, request workflows, and responsible interpretation.
AstroVisor — API, dashboard, and product.
License
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Flicense-qualityDmaintenanceDynamically converts any REST service's OpenAPI specification into MCP tools, enabling interaction with REST endpoints through natural language. Supports Spring Boot services and includes auto-discovery for common API configurations.Last updated
- Alicense-qualityBmaintenanceExposes OpenAPI specifications as MCP tools, enabling AI assistants to explore and understand API structures, endpoints, schemas, and documentation through semantic queries.Last updated24MIT
- Flicense-qualityDmaintenanceDynamically exposes any OpenAPI/Swagger API as tools for AI assistants, automatically generating MCP tools from OpenAPI specs with authentication support.Last updated
- Alicense-qualityDmaintenanceMCP server that provides tools to query 17+ astronomical databases (e.g., SIMBAD, VizieR, Gaia) via HTTP/TAP APIs, enabling AI applications to access astronomical data through natural language.Last updated672BSD 3-Clause
Related MCP Connectors
Free public MCP for AI agents — 193 tools, 44 workflows. No API key.
OCR, transcription, file extraction, and image generation for AI agents via MCP.
Hosted MCP with 91 agent tools: X, domains, SEO, Maps, Trends, Search, YouTube, TikTok, and more.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/rokoss21/astrovisor-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server