# Branchly Muse connector

From musedirectory.ai, the independent directory of Meta Muse connectors. Not affiliated with Meta.

## Branchly

Manage AI agent content, prompts, and actions with built-in analytics and insights

- Record: https://musedirectory.ai/connector/branchly
- Category: Developer Tools
- Developer: Branchly (https://branchly.io)
- Muse status: Extra setup. Not in Muse's Connectors list yet. Muse can still use it: its page gives you a request to paste into Muse.
- Health: Working, 1379ms, checked 2026-09-28T09:01:22Z
- Endpoint: https://api.branchly.io/mcp
- Auth: Needs an access key; Pricing: unknown
- Screening: Screened, no issues found (2026-09-25T00:36:45Z)
- Source: Found in the official MCP Registry (io.branchly/branchly) https://registry.modelcontextprotocol.io/v0/servers?search=io.branchly%2Fbranchly

Branchly connects to your AI agent to manage knowledge bases, configure prompts and tools, and analyze user interactions. Track sessions, sentiment, top searches, and which content gets cited most.

Example request: "Show me which help articles my customers are asking about most and how satisfied they are with the answers"

How to connect: Not in Muse's Connectors list yet, but Muse can still use it. Paste this into Muse: "Use Branchly to help me. It is a free service with an MCP server at https://api.branchly.io/mcp. It needs an API key from Branchly; ask me to enter it through your secure credential prompt. Ask me before you share anything with it." Muse asks before it shares anything with the app's site. Meta does not review apps used this way, so only use ones you trust. We tested this in the Muse app on September 24, 2026: Muse used an app's link directly this way and returned a live answer.

Tools:
- get_application: Return the full configuration of the authenticated application.
- list_nodes: List nodes in the knowledge base. Optionally filter by vertex label(s), data source type(s), or data source IDs. Optionally sort by updated_at ('asc' or 'desc'). Provide 'query' and 'locale' to perform a full-text search across node content (requires both parameters). Returns pag
- create_node: Create a new content node in the knowledge base. Label defaults to 'content'.
- read_node: Read a node by its ID. Returns full node details.
- update_node: Update a content node by its ID. Only provided fields are updated. Title and text are locale dicts (e.g. {"de": "Titel"}) — provided keys are merged, others preserved. custom_metadata keys are likewise merged into the existing metadata.
- list_data_sources: List data sources for the authenticated application. Optionally filter by data source type(s). Returns paginated results ordered by last update.
- create_data_source: Create a new data source for the authenticated application. The 'settings' object must match the given 'type'. Returns the created data source. Creating a data source does not sync any content — use run_data_source to start a sync.
- update_data_source: Update a data source by its ID. Only provided fields will be updated (partial update). Returns the updated data source.
- run_data_source: Trigger an asynchronous run (sync/crawl) of a data source by its ID. The run is queued as a background job. Returns the accepted status. Track progress with read_data_source_runs.
- read_data_source_runs: List data source runs for the authenticated application, latest first. Optionally filter by data source ID or run status. Returns the latest runs (default 5) and the total matching count. Use this to track the progress of a run triggered via run_data_source.
- list_prompts: List prompts for the authenticated application. Optionally filter by type, subtype, or active status. Returns paginated results ordered by active status then last update.
- create_prompt: Create a new prompt version for the authenticated application. The new prompt is automatically set as active and the previously active prompt of the same type and subtype is deactivated. Use this whenever you want to change prompt text.
- update_prompt: Activate or deactivate a prompt by its ID. Use is_active=true to restore a previous prompt version — the currently active prompt of the same type and subtype is deactivated automatically. To change prompt text, use create_prompt instead.
- list_tools: List tools configured for the authenticated application. Optionally filter by active status. Returns all matching tools.
- read_tool: Read a tool by its ID. Returns full tool details.
- create_tool: Create a new tool (AI action) for the authenticated application. Tool names must be unique per application, snake_case, max 64 chars. 'tool_config' and 'function_arguments' must match the given 'tool_type'. Returns the created tool.
- update_tool: Update an existing tool by its ID. Fetches the current tool and applies partial updates to name, description, or active status.
- read_sessions: List sessions for application. Filter by interaction types (chat, search, navigation, form_submission, voice), embed types, answer types, tool IDs, or a full-text search query. Each session includes its full history, but document chunks and tool calls are returned as references (
- read_session_detail: Read a single session with its full history in token-dense form. Returns all interactions (chat, navigation, search, form, voice). Document chunks and tool calls are returned as references (IDs + titles) - use read_node(vertex_id) or read_tool(tool_id) to fetch full content.
- get_active_sessions_over_time: Time series of active sessions, bucketed by day or week (auto-chosen from the time window). Each row breaks the total down by interaction type (chat / search / navigation / form_submission / voice). Gaps are zero-filled, so the series is safe to plot directly. Use this to spot sp
- get_top_locales: Top locales (BCP-47 style, e.g. 'de_DE', 'en_US') across chat, search, and navigation requests, ranked by occurrence count. Use this to understand which languages/regions are being served and where content gaps may exist.
- get_top_languages: Top detected natural languages of user CHAT questions, ranked by occurrence count. Differs from get_top_locales: locale reflects the embed/browser setting, language is detected from the actual query text. Use both to spot mismatches (e.g. German-speaking users hitting an English-
- get_top_devices: Top device categories (e.g. 'desktop', 'mobile', 'tablet') across all request types, ranked by occurrence count. Use this to understand the device mix of real users interacting with the embed.
- get_top_geographies: Top (country, region) combinations across all request types, ranked by occurrence count. Country is an ISO country code; region may be null when unavailable. Use this to understand geographic distribution of users.
- get_sentiment_distribution: Sentiment distribution of chat answers as counts of 'positive', 'negative', and 'neutral'. Optionally restrict to specific answer types. Use this as a quick quality signal — a rising 'negative' share usually warrants drilling into individual sessions via read_sessions.
- get_top_cited_sources: Knowledge-base nodes most frequently cited in chat answers, ranked by citation count. Each row carries the node UUID (vertex_id), its title, optional source URL, and citation_count. Use this to understand which knowledge is actually load-bearing; pair with read_node(vertex_id) to
- get_top_clicked_urls: URLs users actually clicked from inside the embed (search results, citation links, follow-ups, etc.), ranked by click count. Optionally filter by click event type. Use this to see what users find useful enough to click through to.
- get_top_interaction_sources: Page URLs the user was ON when they interacted with the embed (chat, search, navigation, form submission), ranked by occurrence count. Differs from get_top_clicked_urls: this is the ORIGIN page, not the destination. Use this to find which pages of the host site drive the most emb
- get_top_searches: Top user search queries (normalized: lowercased, trimmed), ranked by occurrence count. Use this to discover dominant user intents and content gaps; pair with read_sessions(search_query=...) to inspect specific sessions.
- get_top_tags: Top tags attached to chat answers (auto-derived classifications), ranked by occurrence count. Optionally restrict to specific answer types. Use this for a quick topical breakdown of chat traffic.
- get_answer_type_distribution: Distribution of chat answer types (e.g. 'answered', 'no_answer', 'tool_call', 'human_handoff'), ranked by occurrence count. Use this to monitor answer quality: a high 'no_answer' share signals content gaps; a high 'human_handoff' share signals where the bot is escalating.
- get_trending_classifications: Time series of trending classifications (topics OR intents inferred from chat content), one series per classification id. Each series item carries a period timestamp and count. Use this to see how topical or intent demand shifts over time. Choose 'topic' for subject-matter trends
- read_chat_request_documents: Read the full document chunks retrieved for a single chat request (QA or SA). Returns chunk_id, vertex_id, title, full text, score, source, data source type, page metadata, and whether the chunk was cited in the final answer. Use this after read_session_detail to inspect the exac
- read_chat_request_tool_calls: Read the full tool calls executed for a single chat request. Returns tool_call_id, tool_id, tool_name, tool_type, full arguments (JSON), and full content/response (JSON). Use this to inspect what the assistant invoked and how the tool responded. Ordered by timestamp ascending.
- read_search_request_results: Read the full search results returned for a single instant-search request. Returns chunk_id, vertex_id, title, full text, and relevance score. Use this to inspect the exact results a user saw for their search query. Ordered by score descending.
- get_active_sessions_by_embed: Time series of active sessions broken down by embed type (chat, chat_widget, navigator, search_interface, voice, api). Use this to understand which interfaces are driving usage.
- update_chat_request_analytics: Update analytics fields on a chat request: summary, tags, answer_type, sentiment, classification_topic_id, or classification_intent_id. Only provided fields are updated. Use this to annotate or reclassify chat interactions after the fact.

Screening checks:
- MCP handshake: pass (Answered in 1734ms)
- Domain against threat feeds (Cloudflare security DNS): pass (api.branchly.io, branchly.io not flagged)
- Published packages against the OSV malicious-package database: n/a (No npm or PyPI package published)
- Hidden instructions or invisible characters in tool text: pass (37 tools read, nothing found)
- Inputs asking for passwords, card numbers or seed phrases: pass (None found)
- Domain and redirects: pass (No redirects off the domain)
- AI review of purpose and tool behavior: pass (No concerns)