MCP Reference

Promptmark’s MCP server exposes 74 tools across 12 categories. All tools require authentication via Bearer JWT or API key.

Prompt Tools (13)

list_prompts Prompts

List prompts with optional filtering.

Parameters:

Name	Type	Required	Description
`search`	string	No	Filter by title or content
`tag`	string	No	Filter by tag name
`collection_id`	string	No	Filter by collection ID
`type_id`	string	No	Filter by prompt type
`is_template`	boolean	No	Filter templates only (true) or prompts only (false)
`is_remix`	boolean	No	Filter remixes only (true) or exclude (false)
`limit`	integer	No	Max results (default 50, max 100)
`offset`	integer	No	Pagination offset (default 0)

get_prompt Prompts

Get a specific prompt by ID.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The prompt ID

create_prompt Prompts

Create a new prompt.

Parameters:

Name	Type	Required	Description
`title`	string	Yes	Prompt title
`prompt`	string	Yes	Prompt content (use `{{var}}` for templates)
`notes`	string	No	Private notes
`is_public`	boolean	No	Public visibility (default false)
`is_remixable`	boolean	No	Allow remixing (default false)
`tags`	string	No	JSON array of tags, e.g. `["ai", "coding"]`
`type_id`	string	No	Type: general, system, query, text-to-image, text-to-video, text-to-speech, text-to-code, agent
`collection_id`	string	No	Collection to assign to
`remixed_id`	string	No	Source prompt ID (creates a remix)
`models`	string	No	JSON array of model IDs
`self_rating`	number	No	Rating 1-5 (0 = no rating)

update_prompt Prompts

Update an existing prompt. Only provided fields are changed.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Prompt ID to update
`title`	string	No	New title
`prompt`	string	No	New content
`notes`	string	No	New notes
`is_public`	boolean	No	Public visibility
`is_remixable`	boolean	No	Allow remixing
`tags`	string	No	JSON array of tags
`type_id`	string	No	Prompt type
`collection_id`	string	No	Collection ID (empty string to remove)
`models`	string	No	JSON array of model IDs
`self_rating`	number	No	Rating 1-5

delete_prompt Prompts

Soft-delete a prompt.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Prompt ID to delete

search_prompts Prompts

Search prompts by title, tags, or visibility.

Parameters:

Name	Type	Required	Description
`query`	string	No	Search term for title and notes
`tags`	array	No	Filter by tags (must match ALL)
`is_public`	boolean	No	Filter by visibility
`limit`	integer	No	Max results (default 50)

list_tags Prompts

List all unique tags across your prompts.

Parameters:

Name	Type	Required	Description
`include_counts`	boolean	No	Include usage count per tag (default false)
`limit`	integer	No	Max tags to return (default 100)

get_prompt_schema Prompts

Get the template variable schema for a prompt.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Prompt ID

Returns: Variable names, types, defaults, options, and required status.

render_prompt Prompts

Render a prompt template by substituting variables.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Prompt template ID
`variables`	object	No	Key-value pairs of variable values

Returns: The rendered prompt text with variables substituted.

validate_prompt_inputs Prompts

Validate variable values before rendering.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Prompt template ID
`variables`	object	Yes	Key-value pairs to validate

Returns: Validation errors for invalid or missing required values.

get_prompt_tools Prompts

Get the attached MCP and native tools for a prompt.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	The prompt ID

Returns: Tool references with connection IDs, tool names, and native provider info. Includes count of attached tools.

update_prompt_tools Prompts

Set the full list of attached tools on a prompt. Replaces any existing tools. Each tool reference needs either connection_id + tool_name (for MCP server tools) or native_provider + tool_name + native_type (for native tools like Anthropic web search).

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	The prompt ID
`tool_refs`	array	Yes	Array of tool reference objects (max 50)

Each object in tool_refs has:

Name	Type	Required	Description
`connection_id`	string	No	MCP connection ID (for MCP server tools)
`tool_name`	string	Yes	Name of the tool
`native_provider`	string	No	Provider name for native tools (e.g. `anthropic`, `openai`)
`native_type`	string	No	API type identifier for native tools (e.g. `web_search_20250305`)

remove_prompt_tool Prompts

Remove a single tool from a prompt’s attached tools. For MCP tools, provide connection_id and tool_name. For native tools, provide native_provider and tool_name.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	The prompt ID
`connection_id`	string	No	MCP connection ID (for MCP server tools)
`tool_name`	string	Yes	Name of the tool to remove
`native_provider`	string	No	Provider name for native tools (e.g. `anthropic`)

Collection Tools (6)

list_collections Collections

List all collections.

Parameters:

Name	Type	Required	Description
`include_counts`	boolean	No	Include prompt count per collection (default true)

get_collection Collections

Get a specific collection with optional prompt list.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Collection ID
`include_prompts`	boolean	No	Include prompt IDs (default false)

create_collection Collections

Create a new collection.

Parameters:

Name	Type	Required	Description
`name`	string	Yes	Collection name
`description`	string	No	Description
`icon`	string	No	Icon identifier (default “folder”)
`color`	string	No	Color: gray, red, orange, yellow, green, cyan, blue, purple, pink (default “gray”)
`is_public`	boolean	No	Public visibility (default false)

update_collection Collections

Update a collection. Only provided fields are changed.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Collection ID
`name`	string	No	New name
`description`	string	No	New description
`icon`	string	No	New icon
`color`	string	No	New color
`is_public`	boolean	No	Public visibility

delete_collection Collections

Delete a collection. Prompts in it become uncategorized.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Collection ID

assign_prompt_to_collection Collections

Move a prompt to a collection or remove from its current collection.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	Prompt ID
`collection_id`	string	No	Target collection ID (empty to remove)

Tag Tools (4)

rename_tag Tags

Rename a tag across all prompts that use it.

Parameters:

Name	Type	Required	Description
`old_name`	string	Yes	Current tag name
`new_name`	string	Yes	New tag name

add_tag Tags

Add a tag to a specific prompt. Tags must be alphanumeric with hyphens and underscores, max 50 characters. Duplicate tags (case-insensitive) are ignored.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	The prompt ID to add the tag to
`tag`	string	Yes	Tag name (alphanumeric, hyphens, underscores, max 50 chars)

Returns: Updated tag list and count for the prompt.

remove_tag Tags

Remove a tag from a specific prompt. Uses exact string matching.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	The prompt ID to remove the tag from
`tag`	string	Yes	Exact tag name to remove

Returns: Updated tag list and count for the prompt.

bulk_add_tag Tags

Add a tag to multiple prompts at once. Skips prompts that already have the tag. Uses a database transaction for atomicity.

Parameters:

Name	Type	Required	Description
`tag`	string	Yes	Tag name to add (alphanumeric, hyphens, underscores, max 50 chars)
`prompt_ids`	array	Yes	Array of prompt IDs to add the tag to (max 100)

Returns: Number of prompts updated and total prompts requested.

Remix Tools (1)

list_remixes Remix

List all remixes (prompts derived from other prompts).

Parameters:

Name	Type	Required	Description
`limit`	integer	No	Max results (default 50)

Version Tools (3)

get_prompt_versions Versions

Get version history for a prompt.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	Prompt ID

get_prompt_version Versions

Get a specific version snapshot.

Parameters:

Name	Type	Required	Description
`version_id`	string	Yes	Version ID

restore_prompt_version Versions

Restore a prompt to a previous version. Snapshots the current state first.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	Prompt ID
`version_id`	string	Yes	Version ID to restore

Response Tools (4)

capture_response Responses

Capture an AI response for a prompt.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	Prompt ID
`model_id`	string	Yes	Model name/ID (e.g., `openai/gpt-4o`, `anthropic/claude-sonnet-4`)
`content`	string	Yes	Response content
`metadata`	object	No	Optional metadata (tokens, latency, temperature)

list_captured_responses Responses

List captured responses, newest first.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	No	Filter by prompt ID
`limit`	integer	No	Max results (default 20, max 100)
`offset`	integer	No	Pagination offset

get_captured_response Responses

Get full response content by ID.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Response ID

delete_captured_response Responses

Permanently delete a captured response.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Response ID

Scan Tools (1)

acknowledge_scan Scans

Acknowledge a safety scan issue. Transitions the flag from HAS (1) to HAD (2), allowing the prompt to be published.

Parameters:

Name	Type	Required	Description
`prompt_id`	string	Yes	Prompt ID
`category`	string	Yes	Scan category: `pii`, `injection`, `secrets`, or `danger`

Conversation Tools (11)

list_conversations Conversations

List conversations with optional search and source filtering.

Parameters:

Name	Type	Required	Description
`page`	integer	No	Page number (1-indexed, default 1)
`search`	string	No	Search conversations by title
`source`	string	No	Filter by source: `native`, `claude`, `chatgpt`, `json`, `markdown`

get_conversation Conversations

Get a conversation by ID with optional messages.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The conversation ID to retrieve
`include_messages`	boolean	No	Include messages in response (default true)
`message_limit`	integer	No	Max messages to return (default 50, max 200)

create_conversation Conversations

Create a new conversation with a model and optional system prompt.

Parameters:

Name	Type	Required	Description
`title`	string	Yes	Conversation title
`model_id`	string	Yes	AI model ID for this conversation (e.g. `anthropic/claude-sonnet-4`)
`linked_prompt_id`	string	No	Link to an existing prompt ID for system prompt
`system_prompt`	string	No	Custom system prompt (overrides linked prompt)

update_conversation Conversations

Update conversation metadata (title, model, linked prompt, system prompt). Only provided fields are changed.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The conversation ID to update
`title`	string	No	New conversation title (max 200 chars)
`model_id`	string	No	New model ID
`linked_prompt_id`	string	No	Prompt ID to link (empty string to unlink)
`system_prompt`	string	No	New system prompt text

delete_conversation Conversations

Delete a conversation (soft delete).

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The conversation ID to delete

send_message Conversations

Send a message in a conversation and get the AI response (non-streaming). Requires an API key configured in Settings (supports OpenRouter, OpenAI, and Anthropic).

Parameters:

Name	Type	Required	Description
`conversation_id`	string	Yes	The conversation ID to send a message in
`content`	string	Yes	Message content to send
`prompt_reference_id`	string	No	Optional reference to a prompt ID

import_conversation Conversations

Import a conversation with messages from an external source.

Parameters:

Name	Type	Required	Description
`title`	string	Yes	Conversation title
`source`	string	No	Import source identifier (e.g. `claude`, `chatgpt`)
`model`	string	No	Model ID used in the conversation
`messages`	array	Yes	Array of message objects (see below)

Each object in messages has:

Name	Type	Required	Description
`role`	string	Yes	Message role: `system`, `user`, or `assistant`
`content`	string	Yes	Message content

list_conversation_messages Conversations

List messages in a conversation with cursor-based pagination for large threads.

Parameters:

Name	Type	Required	Description
`conversation_id`	string	Yes	The conversation ID
`before`	string	No	Cursor: return messages created before this ISO 8601 timestamp
`limit`	integer	No	Max messages to return (default 50, max 200)

link_conversation_prompt Conversations

Link a conversation to an existing prompt for system prompt injection.

Parameters:

Name	Type	Required	Description
`conversation_id`	string	Yes	The conversation ID
`prompt_id`	string	Yes	The prompt ID to link

update_conversation_tools Conversations

Replace the full set of attached tools for a conversation. Validates each tool reference against active connections and cached tools.

Parameters:

Name	Type	Required	Description
`conversation_id`	string	Yes	The conversation ID
`tools`	array	Yes	Full list of tools to attach (replaces current set, max 50)

Each object in tools has:

Name	Type	Required	Description
`connection_id`	string	No	MCP connection ID (for MCP tools)
`tool_name`	string	Yes	Tool name
`is_native`	boolean	No	True for provider built-in tools (e.g. Anthropic web search)
`native_provider`	string	No	Provider name for native tools: `anthropic`, `openai`
`native_type`	string	No	API type identifier for native tools (e.g. `web_search_20250305`)

remove_conversation_tool Conversations

Remove a single tool from a conversation’s attached tools.

Parameters:

Name	Type	Required	Description
`conversation_id`	string	Yes	The conversation ID
`connection_id`	string	Yes	MCP connection ID, or `native` for provider built-in tools
`tool_name`	string	Yes	Name of the tool to remove

Playbook Tools (13)

Playbooks are multi-step AI workflows with sequential execution, branching, breakpoints, and elicitation. Content uses markdown with ## STEP, ## BRANCH, ## INPUTS, and ## ARTIFACTS sections. Variables use {{var}} syntax. Steps can capture named outputs with @output(varname) directives and request user input mid-execution with @elicit(type, "prompt").

list_playbooks Playbooks

List playbooks with optional search, tag filtering, and pagination.

Parameters:

Name	Type	Required	Description
`search`	string	No	Search playbooks by title
`tags`	string	No	JSON array of tags to filter by
`page`	integer	No	Page number (1-indexed, default 1)
`page_size`	integer	No	Results per page (default 20, max 100)

get_playbook Playbooks

Get a playbook by ID with its parsed structure (steps, inputs, branches).

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The playbook ID to retrieve

Returns: Playbook metadata, raw content, and parsed definition including steps, input variables, branches, artifact type, and output variables.

create_playbook Playbooks

Create a new multi-step playbook with markdown content containing ## STEP sections.

Parameters:

Name	Type	Required	Description
`title`	string	Yes	The playbook title
`content`	string	Yes	The playbook markdown content with `## STEP` sections
`description`	string	No	Optional description of the playbook
`tags`	string	No	JSON array of tag strings
`default_model_id`	string	No	Default AI model ID for execution
`artifact_type`	string	No	Artifact type for final output: `markdown`, `json`, `mermaid`, `chartjs`, `html_css`, `javascript`, `typescript`. Parsed from `## ARTIFACTS` section if present in content.
`output_vars`	array	No	Informational list of named output variables defined via `@output()` directives in the playbook steps

update_playbook Playbooks

Update an existing playbook. Snapshots the current version before applying changes. Only provided fields are changed.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The playbook ID to update
`title`	string	No	New title
`content`	string	No	New playbook markdown content
`description`	string	No	New description
`tags`	string	No	New JSON array of tag strings
`default_model_id`	string	No	New default AI model ID
`artifact_type`	string	No	New artifact type: `markdown`, `json`, `mermaid`, `chartjs`, `html_css`, `javascript`, `typescript`
`output_vars`	array	No	Informational list of named output variables (read-only, derived from `@output()` directives in content)

delete_playbook Playbooks

Soft-delete a playbook.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	The playbook ID to delete

validate_playbook Playbooks

Validate playbook markdown content without saving. Returns the parsed structure with any warnings or errors.

Parameters:

Name	Type	Required	Description
`content`	string	Yes	The playbook markdown content to validate

Returns: Parsed playbook structure (steps, inputs, branches, artifacts, outputs) or validation errors describing syntax issues.

execute_playbook Playbooks

Execute a playbook with given input values and AI model. Runs each step sequentially with accumulated context. Supports breakpoints to pause execution at specific steps. Returns named output variables from @output() directives and artifact_type from ## ARTIFACTS section. Requires an API key configured in Settings.

Parameters:

Name	Type	Required	Description
`playbook_id`	string	Yes	The playbook ID to execute
`input_values`	object	No	Key-value pairs of input variable values
`model_id`	string	No	AI model ID to use (overrides playbook default)
`breakpoints`	string	No	Comma-separated step numbers where execution should pause (e.g. `"2,4"`). Paused executions can be resumed with `resume_playbook_execution`.

Returns: Execution ID, status (completed, failed, paused, awaiting_input), step results with content and metadata, named outputs, and cost/token totals.

get_playbook_execution Playbooks

Get a playbook execution by ID with all step results.

Parameters:

Name	Type	Required	Description
`execution_id`	string	Yes	The execution ID to retrieve

Returns: Execution metadata, status, per-step results (content, tokens, cost, latency, model), named outputs, and artifact information.

list_playbook_executions Playbooks

List executions for a playbook with pagination.

Parameters:

Name	Type	Required	Description
`playbook_id`	string	Yes	The playbook ID to list executions for
`page`	integer	No	Page number (1-indexed, default 1)
`page_size`	integer	No	Results per page (default 20, max 50)

get_playbook_versions Playbooks

Get version history for a playbook.

Parameters:

Name	Type	Required	Description
`playbook_id`	string	Yes	The playbook ID to get version history for

Returns: List of version snapshots with ID, timestamp, and content summary.

restore_playbook_version Playbooks

Restore a playbook to a previous version. Snapshots the current state first.

Parameters:

Name	Type	Required	Description
`playbook_id`	string	Yes	The playbook ID to restore
`version_id`	string	Yes	The version ID to restore to

resume_playbook_execution Playbooks

Resume a paused playbook execution from the next step after a breakpoint. Optionally override input variables before continuing. The execution must be in paused status from a previous execute_playbook or resume_playbook_execution call.

Parameters:

Name	Type	Required	Description
`execution_id`	string	Yes	The paused execution ID to resume
`variables`	object	No	Optional variable overrides to apply before resuming

respond_to_elicitation Playbooks

Respond to an elicitation prompt in a paused playbook execution. When a step contains an @elicit() directive, execution pauses with status awaiting_input. Provide the response to resume execution from the elicitation step.

Parameters:

Name	Type	Required	Description
`execution_id`	string	Yes	The execution ID awaiting elicitation response
`response`	string	Yes	The user’s response to the elicitation prompt

Skills Tools (6)

User-authored agent skills follow the agentskills.io spec — a single SKILL.md file with YAML frontmatter (name, description, license, compatibility, allowed_tools) and a markdown body. The MCP tools below let an external client manage the skill library (CRUD + search).

Info

Heads up — load_skill is not an MCP tool. Inside a Promptmark conversation, non-draft skills are injected into the system prompt as a sanitized catalog, and the model retrieves full SKILL.md content on demand via a server-injected load_skill tool. That tool only exists inside Promptmark’s own conversation surface — it will not appear in your MCP client’s tool list. See Skills for the in-conversation flow.

list_skills Skills

List skills with optional substring search and pagination.

Parameters:

Name	Type	Required	Description
`search`	string	No	Search by name, title, or description
`page`	integer	No	Page number (1-indexed, default 1)
`page_size`	integer	No	Results per page (default 20, max 100)

get_skill Skills

Get the full SKILL.md for a skill (frontmatter + body). Provide either id or name.

Parameters:

Name	Type	Required	Description
`id`	string	No	Skill ID (UUID)
`name`	string	No	Skill name (max 200 chars)

Returns: The full SKILL.md as a string: ---\n<frontmatter YAML>\n---\n\n<markdown body>.

create_skill Skills

Create a new skill from SKILL.md content. The content is parsed and validated against the agentskills.io spec; safety scans (PII, injection, secrets) run before persistence.

Parameters:

Name	Type	Required	Description
`content`	string	Yes	Full SKILL.md content (YAML frontmatter + markdown body)
`title`	string	No	Optional human-readable display title
`icon_emoji`	string	No	Optional icon emoji (default: sparkle)
`tags`	array	No	List of tag strings (max 50)
`is_draft`	boolean	No	Whether the skill is a draft (default false)

update_skill Skills

Update an existing skill. Snapshots the current state to the version history before applying changes. Frontmatter name and description come from the new content; title, icon_emoji, tags, and is_draft are independent fields.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Skill ID (UUID)
`content`	string	Yes	New SKILL.md content
`title`	string	No	New display title
`icon_emoji`	string	No	New icon emoji
`tags`	array	No	New tag list (max 50)
`is_draft`	boolean	No	Draft state (omit to leave unchanged)

delete_skill Skills

Soft-delete a skill by ID.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Skill ID (UUID)

search_skills Skills

Search skills by name, title, or description substring. Thin wrapper over list_skills that requires a query.

Parameters:

Name	Type	Required	Description
`query`	string	Yes	Search query (max 200 chars)
`page`	integer	No	Page number (1-indexed, default 1)
`page_size`	integer	No	Results per page (default 20, max 100)

MCP Connection Tools (10)

Manage external MCP server connections. Connect to any MCP-compatible server (e.g. Civic Hub, custom servers) and browse their tools from within Promptmark.

list_mcp_connections MCP Connections

List configured external MCP server connections with pagination.

Parameters:

Name	Type	Required	Description
`active_only`	boolean	No	Only return active connections
`limit`	integer	No	Max results (default 20, max 100)
`offset`	integer	No	Pagination offset

get_mcp_connection MCP Connections

Get details of a specific MCP server connection including its available tools.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Connection ID

create_mcp_connection MCP Connections

Add a new external MCP server connection.

Parameters:

Name	Type	Required	Description
`name`	string	Yes	Display name for the connection
`server_url`	string	Yes	MCP server URL (e.g. `https://app.civic.com/hub/mcp`)
`description`	string	No	Optional description
`auth_type`	string	No	Authentication type: `none`, `bearer`, `api_key`, `oauth`, `custom_headers` (default `none`)
`auth_credentials`	string	No	Authentication credentials (token or API key)

update_mcp_connection MCP Connections

Update settings for an existing MCP server connection. Only provided fields are updated.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Connection ID to update
`name`	string	No	Display name
`description`	string	No	Description
`auth_type`	string	No	Authentication type
`auth_credentials`	string	No	Authentication credentials
`is_active`	boolean	No	Whether connection is active

delete_mcp_connection MCP Connections

Remove an MCP server connection and its cached tools.

Parameters:

Name	Type	Required	Description
`id`	string	Yes	Connection ID to delete

refresh_mcp_tools MCP Connections

Refresh the cached tool list for an MCP connection by reconnecting to the remote server and re-fetching available tools.

Parameters:

Name	Type	Required	Description
`connection_id`	string	Yes	ID of the MCP connection to refresh tools for

list_mcp_tools MCP Connections

List cached tools for an MCP connection. Uses locally cached data — no live connection needed.

Parameters:

Name	Type	Required	Description
`connection_id`	string	Yes	ID of the MCP connection to browse tools for

get_mcp_tool MCP Connections

Get details of a specific tool on an MCP connection including its full input schema.

Parameters:

Name	Type	Required	Description
`connection_id`	string	Yes	ID of the MCP connection
`tool_name`	string	Yes	Name of the tool to get details for

list_mcp_executions MCP Connections

List tool execution history with optional filtering by connection. Returns executions ordered by creation date (newest first).

Parameters:

Name	Type	Required	Description
`connection_id`	string	No	Filter by connection ID (omit for all connections)
`limit`	integer	No	Max results (default 20, max 100)
`offset`	integer	No	Pagination offset

favorite_mcp_tool MCP Connections

Set or clear the favorite flag on a cached MCP tool. Favorited tools appear prominently in the tool picker.

Parameters:

Name	Type	Required	Description
`connection_id`	string	Yes	ID of the MCP connection that owns the tool
`tool_name`	string	Yes	Name of the tool to favorite or unfavorite
`is_favorite`	boolean	Yes	`true` to favorite, `false` to unfavorite

PostHog Tools (2)

Configure bring-your-own-key PostHog analytics for AI generation event tracking. Project and personal keys are encrypted with AES-GCM before storage; decrypted keys are never returned by these tools or surfaced in the UI.

configure_posthog PostHog

Configure PostHog analytics with your own project and personal API keys. Validates key prefixes, validates the host URL (HTTPS scheme required, internal/private addresses blocked for SSRF), encrypts the keys, and saves the configuration.

Parameters:

Name	Type	Required	Description
`project_api_key`	string	Yes	PostHog project API key (must start with `phc_`)
`personal_api_key`	string	No	PostHog personal API key for feature flags (must start with `phx_`)
`host`	string	No	PostHog instance URL (default `https://us.i.posthog.com`). Use `https://eu.i.posthog.com` for EU Cloud or your own URL for self-hosted.
`enabled`	boolean	No	Enable or disable analytics (default true)

Returns: Confirmation message with host, enabled, has_project_key, and has_personal_key flags. Decrypted keys are never returned.

get_posthog_status PostHog

Get the current PostHog analytics configuration status. Never returns decrypted keys.

Parameters: None.

Returns: Object with configured, enabled, has_personal_key, and host fields. If no configuration exists, configured and enabled are false and host defaults to https://us.i.posthog.com.