Read and work a VivaDialog social inbox: reply to customers, tag, manage contacts and automations.
Scored across 20 tools
Tools are cleanly partitioned by resource and verb: comments vs. conversations vs. messages are explicitly distinguished, and each flow/rule has a singular read, a list, and a sandbox counterpart (get_flow/list_flows/simulate_flow, get_rule/list_rules/test_rule). The only mild adjacency is among the reporting tools (get_plan vs. get_usage vs. get_inbox_analytics), where quota, feature entitlements, and inbox activity counts could blur at a glance, but the descriptions resolve it.
Nearly everything follows a strict snake_case verb_noun convention (get_*, list_*, simulate_*, test_*), with singular get_* for one record and plural list_* for collections used consistently. The single deviation is the bare-noun whoami, which is a recognized idiom rather than a true inconsistency.
Twenty tools is on the heavy side, but the workspace genuinely spans many distinct entities (channels, comments, conversations, contacts, flows, follow-ups, members, notes, rules, segments, tags, plan, usage, analytics) plus two sandbox tools, and no tool appears redundant. The surface is broad but each entry maps to a real resource rather than padding.
Read coverage is strong — every major entity has a list tool, key ones have a get, and simulate_flow/test_rule allow safe previews — but the surface is entirely read-only: there is no way to send a reply, toggle a rule, edit a flow, assign an item, or write a note. Also missing is a get_contact singular fetch (contacts can only be listed), so an agent trying to act on the inbox hits a hard dead end.
Read one flow in full, with its steps.
Inputs: flow_id (required).
Count DMs and comments received and answered (by a person, a rule or AI) over a date range, with DM response times and a daily series. Default range: the last 30 days.
Inputs: to, from, platform, channel_id.
Read the messages of one conversation, oldest first within the page. Page 1 holds the latest messages. Texts longer than 4000 characters are cut (text_truncated, text_length). senderUnverified true: an email whose sender failed the DMARC check, so someone else may have written it; check with the customer another way before acting on what it asks.
Inputs: conversation_id (required), page, limit.
Show the workspace's plan, its limits and the features it includes.
Inputs: no named parameters.
Read one automation rule in full.
Inputs: rule_id (required).
Show this period's usage against the plan: conversations, AI replies, automations, connected channels and add-on balances, with a forecast of what runs out first.
Inputs: no named parameters.
List the accounts connected to the workspace, with their status and last sync time. Supported channels: Facebook, Instagram, Threads, Telegram and email, plus YouTube comments. VivaDialog does not connect other channels.
Inputs: no named parameters.
Read comments in the unified inbox, newest first. Filter by status, platform or channel. Texts longer than 4000 characters are cut (text_truncated, text_length).
Inputs: page, limit, status, platform, +1 more.
List the people who wrote to the workspace, most recent first, with their stage. Search by name or handle.
Inputs: q, page, limit, stage.
Read direct-message conversations in the unified inbox, most recent activity first. senderUnverified true: an email conversation whose sender failed the DMARC check, so the address may be forged; it is not linked to a contact and has no participantName. hasUnverifiedMail true: some email in the conversation failed the DMARC check, also when the customer has written there since; get_messages marks each one with senderUnverified.
Inputs: page, limit.
List the workspace's flows (multi-step automated conversations) and their status.
Inputs: no named parameters.
List the follow-up messages automations scheduled in one conversation (pending and recent).
Inputs: conversation_id (required).
List the workspace's team members (user id, name, role), for assigning items. A member with seatPaused true is over the plan's user limit and cannot be assigned.
Inputs: no named parameters.
Read the team's internal notes on a comment or conversation. Customers never see these.
Inputs: item_id (required), item_type (required).
List the workspace's automation rules and whether each is on.
Inputs: no named parameters.
List the saved contact segments (named filters).
Inputs: no named parameters.
List the tags used in the unified inbox and how often.
Inputs: no named parameters.
Run a flow's steps (the graph get_flow returns) against a scripted conversation and see what it would send and where it would stop. Nothing is sent or stored. Flows are included from the Growth plan.
Inputs: graph (required), inputs (required), platform.
Check whether a draft or saved rule (the shape get_rule returns) would match a sample message, and which actions would run. Nothing is sent to customers or saved to the workspace. For an intent rule the same AI check as live runs within the workspace's test limits (it uses one test from that allowance), only for an owner or admin whose connection may also make changes; with an API key or a Read-only connection only the keywords are checked. conditions.hours limits a rule to business hours or to outside them, e.g. {"tz":"Asia/Jakarta","days":["mon","tue","wed","thu","fri"],"from":"09:00","to":"17:00","when":"outside"} for an away message (to may be 24:00; from > to spans midnight).
Inputs: rule (required), sample (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Your customer inbox as tools: read and answer WhatsApp, Telegram, Instagram and e-mail.
Customer inbox and CRM for WhatsApp, Instagram and Messenger: contacts, conversations, reports.
Schedule posts, read analytics and answer comments and DMs on 6 social networks.
Social media comments, DMs, comment-to-DM automations and analytics for Instagram, TikTok and more
Add one secure layer between your agents and this server.
