Skip to main content
Glama
Prisme.ai MCP Plugin

Prisme.ai MCP Plugin

Official
by prismeai

Prisme.ai MCP Plugin

Prisme.ai MCP is distributed as a plugin for Claude Code and Codex. The plugin bundles the MCP server, Prisme.ai skills, Claude agents, documentation, and the DSUL linter in one repository. Install and use the plugin only.

What You Get

Component

Description

MCP server

prisme-ai-builder tools for workspaces, automations, apps, events, files, AI Knowledge, and Prisme.ai documentation

DSUL validation

validate_automation, backed by the bundled linter

Skills

/prisme-ai:* skills for connector scaffolding, testing, documentation, fleet sync, A2UI, workspace pages, assistant workflows, and ticket validation

Claude agents

code-review and prisme-assistant for Claude Code

Related MCP server: Ultimate-MCP-Server

Install From GitHub

Repository: prismeai/prismeai-mcp

Claude Code

In Claude Code:

/plugin marketplace add prismeai/prismeai-mcp
/plugin install prisme-ai@prismeai-mcp

Then reload plugins or restart the session if the tools are not visible immediately.

Codex

From a terminal:

codex plugin marketplace add prismeai/prismeai-mcp
codex plugin add prisme-ai@prismeai-mcp

The plugin source is ./plugin inside this repository. Both marketplaces point there, so the same GitHub repo installs cleanly in Claude Code and Codex.

Authenticate

Credentials are user-created API tokens, registered per environment. The recommended path keeps the token out of the chat (it is never sent to the LLM provider):

  1. Create a token in the studio of the target environment: https://<studio-domain>/settings/tokens (e.g. https://sandbox.prisme.ai/settings/tokens).

  2. Run the set-token command in your own terminal — the exact path + config dir are printed in the "no credentials" error:

    node "<plugin>/build/index.js" set-token sandbox --config-dir "<config-dir>"

    It prompts for the token with hidden input, probe-validates it against the API, then saves it to the plugin data dir (credentials.json, mode 600). An invalid token saves nothing.

  3. Re-run your request — the server picks up the new token automatically (no restart). Run set-token again anytime to rotate.

When a tool call has no token (or hits a 401), the error message contains the exact command to run. You can instead let the agent register a pasted token via the set_token tool, but that token is sent to the LLM provider as part of the conversation — prefer the CLI.

First Use

After installation, run:

/prisme-ai:guide

The guide lists every bundled skill and includes the Prisme.ai environment rules, workspace parameter rules, event-search patterns, and recommended workflow.

For environment or token setup help, run:

/prisme-ai:setup

Typical requests:

List automations in ai-knowledge on sandbox
Trace this correlationId in sandbox: <id>
/prisme-ai:app-mcp-implement Salesforce connector

Updating

Pull plugin updates from the marketplace:

Claude Code

/plugin marketplace update prismeai-mcp

Codex

codex plugin marketplace upgrade prismeai-mcp

Release tags rebuild and commit the self-contained bundle at plugin/build/index.js.

Runtime Model

The plugin starts the committed bundle:

plugin/build/index.js

Runtime requirements:

  • Node.js, provided by the host environment

  • No npm install

  • No local build

  • No Playwright

  • No browser token capture

Maintainer Development

Only plugin maintainers need source-based local setup. Use Development to run an MCP client against this repository checkout and rebuild the committed runtime artifact.

Plugin Layout

Path

Purpose

.claude-plugin/marketplace.json

Claude marketplace entry, pointing to ./plugin

.agents/plugins/marketplace.json

Codex marketplace entry, pointing to ./plugin

plugin/.claude-plugin/plugin.json

Claude plugin manifest

plugin/.codex-plugin/plugin.json

Codex plugin manifest

plugin/.mcp.json

Claude MCP server definition

plugin/.codex-plugin/mcp.json

Codex MCP server definition

plugin/build/index.js

Self-contained MCP server bundle

plugin/skills/

Bundled Prisme.ai skills

plugin/agents/

Claude Code agents

plugin/llmDoc/

Prisme.ai documentation exposed to tools

Reference Docs

Guide

Description

Quick Start

Plugin install and first token setup

Manual Setup

Configure non-plugin MCP clients against the committed plugin bundle

Tools Reference

MCP tools exposed by the plugin

Environments

Plugin environment and token persistence

Development

Local repository setup for maintainers and release flow

Available Tools

39 tools
ai_knowledge_completionA

Legacy AI Knowledge API: direct LLM completion without RAG. For new direct model calls, use LLM Gateway v1/chat/completions or v1/embeddings.

Methods:

  • chat: Simple completion using project's configured prompt/model

  • openai: OpenAI-compatible chat completions endpoint

  • embeddings: Generate embeddings for text

  • models: List available models configured in the project

IMPORTANT: Before changing a model in a legacy AI Knowledge project, always call this tool with method='models' first to retrieve the list of available models and verify the model name exists.

Requires a legacy AI Knowledge project API key.

Inputs: apiKey (required), method (required), input, model, +8 more.

ai_knowledge_documentB

Legacy AI Knowledge API: document CRUD operations. For new RAG data, use Storage files/vector_stores APIs.

Methods:

  • get: Get a document by ID

  • list: List documents in a project

  • create: Create a new document (text or URL)

  • update: Update document metadata/content

  • delete: Delete a document

  • reindex: Reprocess a document

  • download: Download original document file

Requires a legacy AI Knowledge project API key.

Inputs: apiKey (required), method (required), projectId (required), id, +15 more.

ai_knowledge_projectA

Legacy AI Knowledge API: project/agent management. For new agents, use Agent Factory /v1/agents APIs.

Methods requiring project apiKey (existing project):

  • get: Get a project by ID

  • update: Update project configuration

  • delete: Delete a project

  • tools: Get available tools for a project

  • datasources: Get available datasources for a project

Methods using user's Bearer token (no apiKey needed):

  • list: List accessible projects

  • create: Create a new project (returns new project with apiKey)

  • categories: List project categories

For methods using Bearer token, use workspaceName/environment to resolve credentials.

Inputs: method (required), ai, id, all, +13 more.

ai_knowledge_queryA

Legacy AI Knowledge API: query an AI Knowledge project with RAG or retrieve context only. For new one-product agents, use Agent Factory messages/send or messages/stream with Storage-backed file_search.

Use method='query' (default) for full RAG response with LLM answer. Use method='context' to retrieve document chunks only without LLM response.

Requires a legacy AI Knowledge project API key (from AI Knowledge > API & Webhooks).

Inputs: text (required), apiKey (required), projectId (required), method, +5 more.

call_apiA

Call any Prisme.ai REST API endpoint, authenticated server-side with the configured environment token (the token is NEVER exposed to the model). Use for endpoints not covered by a dedicated tool — e.g. list organizations ('/orgs'), the current IAM context ('/me'), org members, API keys, etc. path is relative to the environment apiUrl base, which already includes '/v2' (so pass '/orgs', not '/v2/orgs').

Inputs: path (required), body, pick, query, +5 more.

create_automationB

Create a new automation in the Prisme.ai workspace

Inputs: automation (required), workspaceName (required), environment, workspaceId.

create_workspaceB

Create a new Prisme.ai workspace. Returns the created workspace with its ID.

Inputs: workspace (required), environment (required).

delete_automationB
Destructive

Delete an automation from the prisme.ai workspace

Inputs: workspaceName (required), automationSlug (required), environment, workspaceId.

delete_fileB
Destructive

Delete a file from the Prisme.ai workspace storage by id.

Inputs: fileId (required), workspaceName (required), environment, workspaceId.

execute_automationB

Execute/test an automation already existing in the Prisme.ai workspace with optional payload

Inputs: workspaceName (required), automationSlug (required), payload, environment, +1 more.

get_appA
Read-only

Get an app from the Prisme.ai app store with its configuration schema and automations. Use this to understand what config an app requires before installing it in the imports folder. The appSlug is case-sensitive.

Inputs: appSlug (required), environment.

get_app_instanceA
Read-only

Get details of an installed app instance by its slug

Inputs: instanceSlug (required), workspaceName (required), environment, workspaceId.

get_app_instance_configA
Read-only

Get only the configuration of an installed app instance

Inputs: instanceSlug (required), workspaceName (required), environment, workspaceId.

get_automationB
Read-only

Get a specific automation by its slug from the prisme.ai workspace

Inputs: workspaceName (required), automationSlug (required), environment, workspaceId.

get_fileA

Get metadata for a single file by id. Returns the File object including url, name, mimetype, size, metadata.

Inputs: fileId (required), workspaceName (required), environment, workspaceId.

get_prisme_documentationA
Read-only

Returns Prisme.ai documentation by section. Call with 'index' first to see available sections.

SECTIONS:

  • index: Table of contents and quick reference guide

  • automations: Backend logic - triggers, instructions, expressions, memory scopes

  • pages-blocks: UI components - Form, DataTable, RichText, Action, Chat, Charts, Carousel, Tabs, etc.

  • workspace-config: Secrets management, workspace RBAC, one-product IAM notes, native events, versioning with Git

  • advanced-features: Crawler, Custom Code, Agent Factory capabilities, Storage RAG, LLM Gateway, events

  • products-overview: Current one-product platform architecture and integration patterns

  • agent-creation: Agent Factory creation, prompt engineering, RAG, capabilities, evaluations

  • api-selfhosting: REST/webhook API reference, one-product endpoint families, self-hosting deployment

  • product-agent-factory: Agent Factory - agents, publishing, conversations, A2A, tools

  • product-storage: Knowledge (Storage) - files, vector stores, indexing, RAG search

  • product-llm-gateway: LLM Gateway - completions, embeddings, model catalog, routing

  • product-capabilities: Capabilities catalog - MCP, file search, functions, skills, guardrails

  • product-agent-evaluations: Agent Evaluations - test cases, runs, LLM-as-judge

  • product-governance-v2: AI Governance v2 - IAM, API keys, service accounts, observability

  • product-insights-v2: AI Insights v2 - Agent Factory analytics, criteria, feedback, GDPR

  • product-collection-v3: AI Collection v3 - structured data MCP tools for agents

  • product-prompt-library: Prompt Library - MCP prompts and showcases

  • product-builder: Builder - DSUL workspaces, automations, pages, apps

  • capability-workspaces: Backing guardrail, memory, search, vector provider, and connector workspaces

  • legacy-products-overview: Legacy product architecture overview

  • legacy-product-securechat: Legacy SecureChat product details

  • legacy-product-store: Legacy AI Store product details

  • legacy-product-knowledge: Legacy AI Knowledge and Knowledge Client details

  • legacy-product-governance: Legacy AI Governance details

  • legacy-product-insights: Legacy AI Insights details

  • legacy-product-collection: Legacy AI Collection details

Inputs: section.

get_reportsA
Read-only

Retrieve bug reports and feedback submitted about the Prisme.ai MCP tools.

Inputs: page, type, limit, status, +1 more.

install_app_instanceB

Install an app from the Prisme.ai app store into a workspace. Use get_app first to understand the app's configuration schema.

Inputs: appInstance (required), workspaceName (required), environment, workspaceId.

list_app_instancesA
Read-only

List all installed app instances in a Prisme.ai workspace. Returns summary info only (slug, appSlug, appName, disabled). Use get_app_instance for full details.

Inputs: workspaceName (required), environment, workspaceId.

list_appsC
Read-only

Search apps from the Prisme.ai app store

Inputs: page, text, limit, labels, +3 more.

list_automationsA
Read-only

List all automations in the Prisme.ai workspace

Inputs: workspaceName (required), environment, workspaceId.

list_filesB

List files in the Prisme.ai workspace. Supports pagination, sorting and Elasticsearch-style filtering via the query parameter.

Inputs: workspaceName (required), page, sort, limit, +3 more.

publish_appA

Publish a workspace as a new app version in the Prisme.ai app store. On first publish, a slug is required to create the app. Subsequent publishes update the existing app.

Inputs: workspaceName (required), name, slug, description, +3 more.

pull_workspaceB

Download the current workspace from Prisme.ai and extract it to a local directory. This will overwrite existing files.

Inputs: path (required), workspaceName (required), environment, workspaceId.

pull_workspace_versionA
Destructive

Pull a workspace version from a git repository, or roll back/import an existing workspace version.

If the user asks to pull from git, you MUST pass gitPlatform (or repositoryId) — do not call this tool without one of these selectors. Omitting both is only valid when the caller explicitly wants to roll back/import an existing workspace version.

Selector choice:

  • gitPlatform — the default for pulling from git on Prisme.ai. Pass the platform repository id (key under the workspace's platformRepositories). If you don't know the id, pass any plausible value; the tool will reject it and return the list of available platform repos so you can retry. Prefer this selector unless the user specifically names a workspace-level repo.

  • repositoryId — only when the user explicitly references a repo declared on the workspace itself.

Response includes pulledFromGit: true|false — if false after a "pull from git" request, the call was wrong; retry with gitPlatform.

Inputs: versionId (required), workspaceName (required), environment, gitPlatform, +2 more.

push_workspaceA
Destructive

Upload the local workspace directory to Prisme.ai. Creates a backup version before importing. Version name (message) must not exceed 15 characters and only allows letters, numbers, hyphens, and underscores (no spaces).

Inputs: path (required), message (required), workspaceName (required), prune, +2 more.

push_workspace_versionA
Destructive

Push a workspace to a git repository by creating a new version.

If the user asks to push to git, you MUST pass gitPlatform (or repositoryId) — do not call this tool without one of these selectors. Omitting both is only valid when the caller explicitly wants a non-git local snapshot.

Selector choice:

  • gitPlatform — the default for pushing to git on Prisme.ai. Pass the platform repository id (key under the workspace's platformRepositories). If you don't know the id, pass any plausible value; the tool will reject it and return the list of available platform repos so you can retry. Prefer this selector unless the user specifically names a workspace-level repo.

  • repositoryId — only when the user explicitly references a repo declared on the workspace itself (under repositories: with mode read-write).

Response includes pushedToGit: true|false — if false after a "push to git" request, the call was wrong; retry with gitPlatform.

Common errors:

  • 400 "not up-to-date": the remote has newer commits — pull the workspace and retry.

  • A successful push holds a ~30 min write lock on the workspace; use unlock_workspace to release it early.

Inputs: description (required), workspaceName (required), name, environment, +3 more.

report_issue_or_feedbackA
Read-only

Report bugs or feedback about the Prisme.ai MCP tools.

Use PROACTIVELY when you encounter issues - even mid-task. Don't wait for task completion.

Trigger on: tool errors, misleading documentation, or missing constraint information.

IMPORTANT: Report how to improve MCP tool guidance, NOT requests to change API behavior.

  • GOOD: "Document that 'message' field is limited to 15 characters"

  • BAD: "Increase the message limit to 50 characters"

Inputs: type (required), message (required), context.

search_eventsA
Read-only

Search for events in Prisme.ai workspace using Elasticsearch DSL.

EVENT STRUCTURE:
Events contain the following key fields:
- @timestamp: Event timestamp (ISO 8601 format) - USE THIS FOR SORTING, NOT "timestamp"
- id: Unique event ID
- type: Event type (e.g., "runtime.automations.executed", "workspaces.pages.updated", "error")
- source: Metadata object containing:
  - correlationId: Groups all events from a single API request/operation
  - userId: User who triggered the event
  - sessionId: User session identifier
  - workspaceId: Workspace identifier
  - automationSlug: Automation name (for automation-related events)
  - http: HTTP request details (method, path, hostname, ip)
  - host: Service information (replica, service name)
- payload: Event-specific data (varies by event type)
- createdAt: Creation timestamp

COMMON QUERIES:
- Find all events for a specific request: {"bool": {"filter": [{"term": {"source.correlationId": "uuid-here"}}]}}
- Find automation executions: {"bool": {"filter": [{"term": {"type": "runtime.automations.executed"}}]}}
- Find events for specific automation: {"bool": {"filter": [{"term": {"source.automationSlug": "automation-name"}}]}}
- Find errors: {"bool": {"filter": [{"term": {"type": "error"}}]}}
- Exclude specific correlationId: {"bool": {"must_not": [{"term": {"source.correlationId": "uuid-here"}}]}}

SORTING:
- Always use "@timestamp" field for time-based sorting: [{"@timestamp": {"order": "desc"}}]
- DO NOT use "timestamp" as it's not mapped in the index

COMMON EVENT TYPES:
- runtime.automations.executed
- runtime.interactions.triggered
- runtime.dsul.updated
- workspaces.automations.created/updated/deleted
- workspaces.pages.created/updated/deleted
- error

ADDITIONAL INFO:
- SearchError are probably caused by your own previous failed attempt to create filters. This is usually not the event the user ask for. Keep searching for the events before the SearchError ones.
- Always use "source" to filter only the necessary fields. Only include the informations that are relevant to the user's request. Ignore durations, timestamps, IP adress if not asked for.

Supports full Elasticsearch DSL query syntax including aggregations, sorting, and pagination.

Inputs: query (required), workspaceName (required), aggs, page, +6 more.

search_workspacesA
Read-only

Search for workspaces by name, description, or slug. Returns workspace IDs and names. Use this to find a workspaceId from a text search.

Inputs: environment (required), name, page, slug, +3 more.

set_tokenA

Register (or rotate) a user-created Prisme.ai API token for an environment. PRIVACY: calling this tool means the token travels through the conversation and is sent to the LLM provider. PREFER the out-of-band CLI instead — tell the user to run node <plugin>/build/index.js set-token <environment> --config-dir <dir> in their own terminal (the exact command is included in the 'no credentials' error); it prompts for the token with hidden input and never exposes it to the chat. Only use this tool if the user explicitly chooses to paste the token here despite that. The token is validated with a probe call to the API before being persisted to the MCP config dir (credentials.json, mode 600); an invalid token persists nothing. Pass apiUrl (and optionally studioUrl) to register an environment that is not configured yet.

Inputs: token (required), environment (required), apiUrl, studioUrl.

uninstall_app_instanceB
Destructive

Uninstall an app instance from a workspace

Inputs: instanceSlug (required), workspaceName (required), environment, workspaceId.

unlock_workspaceA
Destructive

Remove the write lock from a Prisme.ai workspace. Use this when a workspace is stuck in a locked state.

Inputs: environment, workspaceId, workspaceName.

update_app_instanceB

Update an installed app instance (config, disabled status, labels)

Inputs: appInstance (required), instanceSlug (required), workspaceName (required), environment, +1 more.

update_app_instance_configB

Update only the configuration of an installed app instance

Inputs: config (required), instanceSlug (required), workspaceName (required), environment, +1 more.

update_automationB
Destructive

Update an existing automation on the prisme.ai workspace

Inputs: automation (required), workspaceName (required), automationSlug (required), environment, +1 more.

update_reportA

Update an existing bug report or feedback.

Use this to cancel a report, edit its message, or change its type.

Inputs: reportId (required), type, status, message.

upload_fileA

Upload a file to the Prisme.ai workspace's file storage. Returns the uploaded File object(s) with url, id, name, mimetype, size. The file source can be provided in one of three ways: path (local filesystem), url (remote URL fetched by the MCP), or dataUri (data:mime;base64,... string). Exactly one of these must be set.

Inputs: workspaceName (required), url, path, public, +8 more.

validate_automationA
Read-only

Validate Prisme.ai automation(s). Checks schema compliance, expression syntax ({{variables}} and {% code %}), unknown functions, naming conventions, and optionally strict mode. Accepts a file path, folder path (validates all .yml/.yaml/.json files), or automation object. Returns warnings (e.g., missing arguments declaration or naming convention issues) alongside errors even when the automation is valid.

Inputs: path, strict, automation, validateExpressions.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 39 tools

    Disambiguation4/5

    Most tools have distinct purposes, with clear naming distinguishing between operations like create, delete, get, list, update for different resources (automations, files, app instances). There is a slight potential for confusion between similar tools like get_app_instance and get_app_instance_config, but descriptions clarify. Overall, the set is well-organized and an agent can discriminate effectively.

    Naming Consistency5/5

    All tools follow a consistent verb_noun pattern using snake_case (e.g., create_automation, delete_file, get_app). No mixing of conventions. Even the legacy AI tools use a consistent 'ai_knowledge_' prefix followed by a noun. This high consistency aids predictability.

    Tool Count2/5

    With 39 tools, this server is over the typical well-scoped range. While each tool has a specific purpose, the number is overwhelming for an agent, especially given that some domains (e.g., app instances) have many specialized tools that could potentially be merged. This number would better fit a more focused scope.

    Completeness4/5

    The server covers CRUD for automations, files, app instances, and reports, plus workspace management, event search, documentation, and a generic API call tool. However, there is no explicit tool for deleting a workspace, which is a notable gap. The generic call_api can likely handle it, so it's a minor gap. Overall, the tool set is fairly complete for the platform's core operations.

    Related MCP Connectors

    Related MCP Servers