← Plugin catalog
Business & Operations

Follow Up Boss

Peter Yao v0.2.5

Your real estate CRM assistant

Create a free account to save this plugin.

From the developer

Declared capabilities

  • Find contacts and review recent notes
  • Review tasks, appointments, and smart lists
  • Prepare SMS and email drafts

Cost & access

Uses an existing Follow Up Boss account. The plugin does not support purchases, payments, or upgrades.

Declared limitations

  • N1 — General CRM advice. Fresh chat, FUB installed and available for automatic selection, with no forced @FUB mention. No fixtures required. Expected: Explain the general concepts directly. No FUB discovery, authentication prompt, account read, widget or write. CRM terminology alone does not require private account access.
  • N2 — Editing supplied outreach text. Fresh chat, FUB installed and available for automatic selection, with no forced @FUB mention. No fixtures required. Expected: Rewrite only the supplied text. Do not search for Taylor, retrieve notes, invoke FUB drafting tools, save a draft or send anything. The supplied text is sufficient.
  • N3 — Property search outside this integration. Fresh chat, FUB installed and available for automatic selection, with no forced @FUB mention. No fixtures required. Expected: Do not invoke any FUB tools or search CRM contacts as if they were property listings. Use an appropriate property-search capability if independently available; otherwise explain the need for current listing data. Do not fabricate listings or availability.
Sources & dates

Developer declarations; not independently tested. Listing last checked Nov 18, 2026. Package observed Nov 18, 2026.

View package source →

Description

From the marketplace listing

Follow Up Boss helps real estate agents and teams work with their CRM in chat. Connect your existing Follow Up Boss account to review contacts and recent notes, plan follow-up from tasks, appointments and smart lists, draft personalized outreach, and analyze agent activity and conversation counts. Prepare supported CRM changes for review and confirm them in the review interface before saving. Email and SMS messages remain drafts; the plugin does not send messages or place calls. Requires Follow Up Boss sign-in and applicable account permissions. Results reflect the records and coverage available to your connected account.

Language: English · Automatically detected from descriptions.

Changes

No changes recorded yet. Follow this plugin to keep track of future changes.

First seen Nov 18, 2026. Last seen Nov 18, 2026.

Similar plugins

Similar plugins could not be prepared right now. Please try again in a moment.

Browse plugins →

Keywords

Files & skills

File archives

Plugin package265 files · 174 KBBrowse files →
Skill instructions
fub-crm-updates5.42 KB

View saved version →

---
name: fub-crm-updates
description: Prepare reviewed Follow Up Boss changes to contacts, notes, tasks, appointments, deals, assignment, call logs, saved email drafts or action-plan pauses, and verify the confirmed outcome.
---

# FUB CRM updates

Use the user's requested scope and the live FUB tool definitions. Resolve exact records and account vocabulary first. See [tool contracts](../fub-mcp-tools/references/tool-index.md) for required fields and limits. Never infer a successful write from a prepared review.

## Choose the review workflow

| Requested change | Preparation | Widget-controlled apply |
| --- | --- | --- |
| Add one contact | `create_contact` with exactly one row | `apply_create_contact` |
| Note, stage, profile/contact detail, relationship or new follow-up task | `interaction_update` | `apply_interaction_update` |
| Meeting summary for one identified person | `draft_meeting_note`, then its official interaction review | `apply_interaction_update` |
| Complete, reopen or reschedule an existing task | `get_tasks` → `render_task_review` for exactly one ID | `apply_task_update` |
| Create, edit, cancel or record an appointment outcome | `appointment_update` | `apply_appointment_update` |
| Create or update one deal | `render_deal_review` | `apply_deal_update` |
| Owner, co-agent, lender or pond change for one contact | `render_assignment_review` | `apply_contact_assignment` |
| Log a call already made | `log_call` | `apply_log_call` |
| Pause one action-plan enrollment | `pause_action_plan` | `apply_pause_action_plan` |
| Save an email draft | `send_message` | `apply_save_email_draft` |
| Delete one note, task or deal | `review_delete_record` | `apply_delete_record` |

Apply tools are documented for understanding receipts and host integrations. They are app-only: let the rendered widget collect explicit confirmation and submit the frozen proposal. Do not manually obtain tokens, call apply from chat, or set a confirmation boolean on the user's behalf. If the host cannot render or operate the review, explain the blocker and provide the supported FUB handoff.

## Prepare concrete changes

- Select exact IDs from authorized reads. Use `get_account_schema` or relevant metadata readers for account-specific stages, pipelines, users, timeframes and targets. An unreturned field is not evidence it is blank.
- Contact creation uses one row and stable `rowId`. Exact email/phone duplicates reconcile to existing records; near-name matches need the supported resolution. Do not silently merge or force duplicates.
- New tasks belong in `interaction_update` with a unique `taskRequestId`, due date/time, and timezone as needed. Existing tasks belong in `render_task_review` with a unique `requestId` and exactly one `taskId` in `taskIds`. Do not replace a task to simulate an update.
- Deal and assignment reviews also use a unique request ID for the intent. Carry explicit pipeline, stage, owner and target IDs. A projected close date needs a timestamp with an explicit offset resolved in the user's timezone.
- Contact assignment supports owner, coagent_add, coagent_remove, lender and pond. It does not reassign existing tasks. Preserve collaborator lists through the official proposal; do not simulate routing as a relationship edit.
- Appointment timestamps require explicit offsets. Resolve ambiguity in contact or appointment selection before another preparation call. Appointment changes do not send invitations; the user sends them from FUB.
- Call logs describe a call that already happened; they never dial. A refused incoming missed-call outcome must not be rewritten into a different outcome merely to pass validation.

## Confirm and verify

Show the actual rendered target, old/new values and side effects. Let the user confirm each review; one confirmation applies to one contact, task, deal, delete or enrollment where that workflow requires it. User approval of a larger goal does not turn one signed proposal into approval for extra record

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 1 — file references are metadata, not package contents.

fub-follow-up5.73 KB

View saved version →

---
name: fub-follow-up
description: Plan a Follow Up Boss user's day, find contacts and follow-up leads, inspect notes and smart lists, and draft personalized outreach from FUB data.
---

# FUB follow-up

Check live tool availability using [FUB MCP tools](../fub-mcp-tools/SKILL.md). Read exact inputs from its tool reference when needed; do not assume every capability is enabled.

## Route the request

| User wants | Tool / sequence |
| --- | --- |
| My morning brief or what's left today | `get_my_daily_brief`; set `remainingOnly` for the remainder |
| Tasks only | `get_tasks` with the matching planning view |
| Schedule today, tomorrow or a date range | `get_appointments`; optionally `render_appointment_calendar` with its unchanged calendar result |
| Find a person and show a card | `find_contact` |
| Search/filter contacts for analysis | `query_people`; optionally `render_people` with exact returned people |
| Full details for a resolved contact | `query_people` with `personId` alone |
| Identify the contact in a recording or note | `identify_contact`; wait for `confirmed.personId` |
| Recent notes for an exact person | `list_notes` |
| Search note text | `search_notes` |
| Who needs a call today | `get_leads_needing_followup` |
| Newly created leads with no exposed outreach | `get_unworked_new_leads` |
| Inbound messages without a later exposed reply | `get_messages_by_response_status` |
| Declining engagement | `get_declining_engagement_leads` |
| Old contacts with note-based reengagement context | `get_reengagement_candidates` |
| Zillow leads at risk | `get_at_risk_zillow_leads` |
| Leads available to claim | `get_claimable_leads`; claiming happens in FUB |
| Smart-list names, counts, visible filters | `query_smart_lists` |
| People in a smart list | `get_smart_list_people`; use its continuation; `render_smart_list_people` for a curated exact-ID subset |
| Broad scored outreach ideas | `get_outreach_opportunities`, only if available |
| Draft outreach for selected people | `draft_outreach_messages` with authoritative `personIds`; check each draft against its context and use `send_message` for each draft needing personalization |
| One SMS/email or template-based draft | `send_message` |
| Contact's phone/call link | `call_contact`; the user places the call |

Do not substitute one cohort definition for another. A new lead with no outbound activity, a lead overdue for follow-up, and a lead with no later exposed response answer different questions. If the requested tool is unavailable, explain the limitation before offering a narrower available result.

## Work from canonical records

Resolve people by name, email or phone, and let the user choose among ambiguous matches. For a recording, use up to four mentioned names and any supported appointment evidence; ranked candidates alone do not identify the person. Do not infer contact IDs, lead source, channel preferences or unreturned fields.

Carry returned cursors and coverage warnings through subsequent pages. `query_people` combines different filter types with AND; supplied tags match any supplied tag. When the request requires all supplied tags, verify the intersection from returned authoritative data or state the coverage limit.

`render_people` accepts 1–25 canonical records; `selectedPersonId` must be one of them. Separate suggested reasons or next steps from source fields. `list_notes` already renders a timeline; summarize the useful context instead of repeating entire notes.

## Planning and outreach

Planning dates use the signed-in FUB user's timezone unless the user specifies another. Preserve all-day dates and tasks whose contact is missing. `get_tasks` planning pagination uses its returned cursor; page group totals are not account totals.

FUB appointment data does not prove availability: external calendars and recurrence may be absent, and incomplete coverage suppresses candidate gaps. Say “no recorded FUB appointment in the returned coverage” when supported, n

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 1 — file references are metadata, not package contents.

fub-mcp-tools4.61 KB

View saved version →

---
name: fub-mcp-tools
description: Discover and use Follow Up Boss MCP tools, inspect their arguments and results, connect a FUB account, or diagnose missing tool access. Use with FUB CRM workflows when tool selection or schema details are needed.
---

# FUB MCP tools

Use the connected Follow Up Boss MCP for account data and supported actions. The production endpoint is `https://fub-mcp.zillow.com`; sign-in belongs in the host's OAuth flow. This skill carries no credentials and does not grant FUB permissions.

## Discover before calling

1. Inspect the tools actually exposed by the current connection. Load deferred definitions when supported. Use their exact names, input schemas and instructions. The bundled [tool index](references/tool-index.md) is a source snapshot, not proof that all tools are deployed or enabled.
2. Canonical source names in the reference may appear with a host prefix or gateway alias. In particular `get-current-user` may be normalized by the host. Match live descriptions and schemas; do not mechanically construct names or call a missing tool through another API.
3. If the connection exposes only `request_team_access`, call it without `confirm` for a requested CRM task and relay the access status. `confirm: true` notifies the owner; use it only after the user explicitly agrees. Do not interpret restricted discovery as an empty CRM account.
4. Resolve the signed-in profile with the available current-user tool when identity, scope or timezone is needed. Use returned IDs for contacts, tasks, users, teams, stages and pipelines. Preserve each input's ID type; some tools accept strings and others integers.
5. Optional `intent` is one short description of the task, without names, email addresses, phone numbers or message text. It is not a place to copy user or CRM content.

## Read only the needed reference

- [Tool index](references/tool-index.md): all documented tools, availability caveats, purpose and links to individual contracts.
- `references/tools/<canonical-name>.md`: exact description, required arguments, behavior and links to input/output JSON Schemas for one tool. These schemas include nested fields and limits; a top-level optional property can still be conditionally required by the handler.
- [Call examples](references/call-examples.md): illustrative requests and data-to-render handoffs. Replace fixture IDs with authoritative live results before use.
- [Read-only operation catalog](references/read-operations.md): `fub_reference` and `fub_execute`, when available. These are named GET operations, not arbitrary code execution or write access.
- [Troubleshooting and feedback](references/troubleshooting.md): auth, permissions, partial results, uncertain writes and the distinct feedback paths.

For a local lookup without loading the full catalog:

```sh
python3 scripts/tool_reference.py --list
python3 scripts/tool_reference.py get_tasks --part input
python3 scripts/tool_reference.py --search "appointment"
python3 scripts/tool_reference.py --op tasks.list
```

Paths above are relative to this skill directory. The script reads bundled JSON only and makes no network calls.

## Preserve the tool's contract

- Pick the purpose-built workflow when it matches the request. Use raw readers for underlying records, and renderers only with the producer's canonical result. Preserve IDs, order, timestamps, coverage and error fields; put commentary in separate annotations.
- Resolve ambiguity before a contact-specific action. A candidate list from `identify_contact` is not confirmation: wait for `confirmed.personId` or an explicit selection.
- Continue with the tool's returned cursor/offset and unchanged filters. Distinguish partial, unavailable, forbidden and empty. Do not convert a bounded sample into an account total.
- Creation, modification, due date, occurrence date and activity date are different filters. Use the user's FUB timezone for local planning and explicit offsets for timestamp bounds.
- Review tools pre

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 251 — file references are metadata, not package contents.

fub-reporting4.64 KB

View saved version →

---
name: fub-reporting
description: Analyze Follow Up Boss team performance, agent activity, conversation counts, follow-up queues, lead sources, stage consistency and bounded CRM records while preserving metric definitions and coverage.
---

# FUB reporting

Inspect the live FUB tools before choosing a report. The [catalog](../fub-mcp-tools/references/tool-index.md) documents the source surface; deployment and user permissions may expose fewer tools. Prefer purpose-built reports to reconstructing metrics from incomplete raw lists.

## Choose by metric

| Question | Tool |
| --- | --- |
| Team performance, response speed, trends, stuck stages | `get_team_performance` |
| Manager daily brief and per-agent next actions | `get_team_daily_brief` → `render_team_daily_brief` |
| Follow-up leads for the selected brief agent | `get_agent_follow_up_leads` → `render_agent_follow_up_leads` |
| Internal manager note for the selected lead | `draft_manager_follow_up_note` from the canonical selected lead; posting uses the separate interaction review |
| Per-agent counts under the follow-up rule | `get_needs_follow_up_counts` |
| Calls and occurred appointments by agent | `get_agent_activity_counts` |
| Qualifying call conversations for one agent | `get_agent_conversation_count` |
| Qualifying call conversations for a supplied team | `get_team_conversation_counts` |
| Bounded lead-source comparison | `get_source_performance`, if available |
| Stage/timeframe/note contradictions | `get_stage_definitions` then `get_stage_consistency_review` |
| Why a contact appears in a list or automation | `inspect_fub_diagnostics` plus relevant authorized readers |
| Account vocabulary and completeness | `get_account_schema`; select only required resource groups |

For raw data, use the appropriate `get_events`, `get_calls`, `get_text_messages`, `get_tasks`, `get_appointments`, `get_deals`, `get_pipelines`, `get_users`, `get_teams`, `get_automations`, `get_action_plans`, `get_custom_fields`, `get_templates` or `get_relationships` contract. `query_people` supplies contact searches/details; `list_notes` and `search_notes` supply note context. Read each schema rather than transferring filters between resources.

## Keep definitions explicit

- Qualifying conversations in the conversation-count tools are active public calls of at least 120 seconds attributed by `userId`, not all messages or all attempted calls. Supply authoritative agent IDs; for the team tool preserve exact ID/name pairs.
- `get_agent_activity_counts` reports occurred appointments, not appointments created by the agent. Account-wide sent-text counts are unavailable there; null is not zero.
- `get_team_performance` uses the FUB report definitions. The team mean response time is the average of agents' averages, while median/slowest use the team's underlying leads. Lead-stage snapshots are current snapshots, not date-window changes. Preserve returned exceptions.
- `get_source_performance`'s `dealsPerScannedPerson` is a ratio of unwindowed first-page deals to activity-window people; it can exceed 1 and is not a conversion rate. Do not relabel it as one.
- Stage consistency uses `updated` as a proxy because native stage-change time is unavailable. Explain this limit rather than claiming exact days since the stage changed.
- `get_account_schema` distinguishes available, empty, forbidden and unavailable resources. Account tag vocabulary is unavailable; tags can be read from contacts. Template identity is not template content.

## Bound conclusions by the evidence

Choose explicit dates and timezone matching the user's question. Record-created filters do not measure occurrence, completion, close or activity dates. Use unchanged filters with returned pagination. Raw call phone filtering is local to a source page, so an empty matching page does not establish no calls exist.

Keep totals as lower bounds when the tool says coverage is incomplete; suppress unavailable totals rather than calculating false 

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 1 — file references are metadata, not package contents.

Release notes

Adds five positive and three negative review cases and the verified demo recording. Reconciles test instructions with existing sample data: current-day brief checks, current latest notes and uniquely named review writes. Preserves the four skills and production MCP connection. Test definitions are not claims of passed execution.

Declared in the saved package. Remote tools may change independently.

Package details

Publisher declarations from the archived package. These are separate from our research and the live service's terms.

Package author
Peter Yao
Keywords
See keywords
Declared availability
No country restrictions declaredDeclared regions; live availability may differ.
Commerce declaration
Does not support commerceThis does not establish whether access is free or paid.
Review scenarios
5 positive · 3 negativeScenario counts, not test results.

Declared capabilities

  • Find contacts and review recent notes
  • Review tasks, appointments, and smart lists
  • Prepare SMS and email drafts
  • Review supported CRM updates

Package observed Nov 18, 2026.

Technical details
First seen
Nov 18, 2026 · 06:00 UTC
Last seen
Nov 18, 2026 · 06:00 UTC
Collection status
Collected

plugin_asdk_app_6aecf5792cf48191b788d39e9924766e

Download plugin data (JSON)

Plugin details: freshly prepared. Prepared . Refreshed when plugin data changes or after 6 hours. Your account and Follow status are live.