MCP App Store
by KeenEthics

Overview

YCH Assistant helps authorized YCH users review their own current activity, task categories, goals, production summary, and development priorities, identify the selected connected YCH profile, and search approved YCH/GE knowledge. The prelaunch Plugin is read-only. It does not access named clients, cases, meetings, or calendars, and it cannot create or change YCH business records.

Tools

accept_lead_offer

ChatGPT
Accept a pending YCH lead offer for the authenticated caller only after the user explicitly chooses and asks to accept it. Use an assignment_id from get_my_lead_offers; never send contact or actor identifiers.

confirm_person_memory

ChatGPT
Ask the connected human to approve the exact current Memory proposal displayed by YCH. Approval is accepted only from this same MCP session as an elicitation result with action 'accept' and exactly {approve:true}; model text or arguments cannot approve. This tool accepts only a memory_id and fails closed if hosted human confirmation is unavailable.

confirm_prepared_task_action

ChatGPT
Confirm only the exact task operation already retained by this MCP session. Inputs are opaque operation and confirmation identifiers; this tool exposes no action selector and accepts no task or mutation arguments. It requires MCP form elicitation and protocol-level human acceptance. Model text never authorizes confirmation.

create_calendar_event

ChatGPT
Create an ordinary personal Google Calendar event for the authenticated caller after the user has explicitly asked for it. YCH creates and durably binds the trusted invocation identity internally before dispatch; never ask the user or model to provide an operation_id, operation_key, invocation_id, or tool-call authority field. After UNKNOWN, refresh Calendar and do not issue a replacement identical write; YCH keeps the unresolved invocation bound to its existing operation. A genuinely new settled tool invocation remains a new operation. Client resolution is optional and this tool accepts no client/person/prospect IDs. If the user supplied a date/day such as today but no start time, do not call this tool: ask only What time today? or the equivalent missing-time question. Never invent the user's start time. Once a start time is supplied, unresolved names or nicknames may remain plain event text and must not block creation. A fixed 30-minute product default is used only for event duration when no duration is supplied.

create_crm_note

ChatGPT
Append a note to the selected Person's CRM history only when the user explicitly asks to add that note. Use a resolution_handle from find_my_client or get_my_client; never provide a Person, prospect, profile, or agent ID. Reuse the same operation_key with the exact same note to reconcile an uncertain result. Notes do not create tasks, interactions, follow-ups, or messages.

create_task

ChatGPT
Create a task for the authenticated caller only when they clearly asked for it. A caller-stable UUID operation_key is required and must be reused with the exact same fields to reconcile after UNKNOWN; never mint a new key. Types follow_up and reminder create ordinary tasks; they do not deliver reminders. This does not attach a client or case, send a message, or change an existing task. Include a timezone in due_at.

execute_confirmed_task_action

ChatGPT
Execute only the exact server-retained operation tied to this same-session confirmation. Inputs are opaque operation and confirmation identifiers; this tool exposes no action selector and accepts no task or mutation arguments. A replay with the same identifiers may reconcile the existing durable receipt; UNKNOWN never authorizes a replacement mutation.

find_my_client

ChatGPT
Find an authorized client for the authenticated caller using Person360 across YCH CRM and synchronized GE data by natural name/query or an exact phone/email where allowed. Use this for generic find-my-client requests. Do not use legacy prospect-only lookup. If multiple candidates are returned, ask the user to choose one. The result contains actor-bound opaque resolution handles instead of raw Person IDs.

get_approved_ge_knowledge

ChatGPT
Search only Howard-approved current YCH/GE sources for a focused factual question about a product, process, or organizational guidance. The backend fails closed while its knowledge gate or Howard owner identity is unavailable; withheld, unapproved, stale, or conflicting evidence is not grounded. Do not include client names, identifiers, or contact details. Empty or unavailable results mean approved current evidence did not establish the answer.

get_coach_context

ChatGPT
Read the authenticated caller's own YCH development priorities and goal summary. Use when they ask what skill to improve or want coaching grounded in their YCH goals. It does not return client, case, meeting, calendar, or cause-of-performance data; suggestions remain suggestions.

get_my_calendar

ChatGPT
Read the authenticated caller's personal calendar from today through the next few days (7 by default, at most 14). Returns event times and bounded status/type fields only, with completeness disclosure; it omits event titles, locations, attendees, and client details.

get_my_capability_profile

ChatGPT
Read the authenticated caller's recorded preferences, confirmed skills, observed activity, production, eligibility, availability, and known gaps. This read does not score or rank the agent.

get_my_challenge_squads

ChatGPT
Read the authenticated caller's current YCH challenge squad summaries. Ranking and scores are included only when the canonical result says ranking is available; never infer missing rankings. This tool is read-only and accepts no actor or record identifiers.

get_my_client

ChatGPT
Read a bounded Person360 summary and client brief for one client selected from find_my_client. Accept only the actor-bound resolution_handle returned by YCH; never accept or invent a profile, agent, Person, prospect, or GE client ID.

get_my_context

ChatGPT
Read the authenticated caller's own current YCH task IDs, categories, statuses and due times, primary goal, top development skills and year-to-date production. Use a task_id from this owner-scoped read for a task action. Components YCH cannot currently provide are omitted and the result is marked degraded. It does not return task titles, client details, ranked work or appointment verification.

get_my_day

ChatGPT
Read the authenticated caller's ranked YCH day with who, why, when, and next steps. It preserves YCH's order, shows bounded client context only when authorized, and does not perform actions.

get_my_ge_production

ChatGPT
Read the authenticated agent's own stored GE production summary and freshness. Stale data is not current.

get_my_ge_servicing_update

ChatGPT
Read one GE servicing update using its event_id handle from an authorized source.

get_my_ge_servicing_update_counts

ChatGPT
Read current-freshness GE servicing update counts; counts are withheld unless the source is current.

get_my_lead_offers

ChatGPT
Read only the authenticated caller's owner-scoped pending lead offers needed to choose an offer. YCH keeps contact details locked until acceptance.

get_my_notifications

ChatGPT
Read up to 10 current notifications for the authenticated caller from YCH. The result omits notification and entity IDs, keeps only bounded text/status/timestamps, and redacts common contact and identity-number patterns. This tool is read-only and accepts no actor or record identifiers.

get_my_team_goals

ChatGPT
Read up to three current goals for the authenticated caller's YCH team. YCH remains authoritative for team membership, progress completeness, and status; incomplete progress is returned as pending_data. This tool is read-only and accepts no actor or record identifiers.

get_profile

ChatGPT
Return one stable opaque identity and a synthetic label for the YCH profile authenticated by this connection. This is a read-only identity check and accepts no account selector.

list_my_ge_non_face_to_face_summaries

ChatGPT
List bounded Great Eastern non-face-to-face (NFTF) summaries for the authenticated agent, with explicit freshness and completeness.

list_my_ge_policies

ChatGPT
List bounded persisted GE policy summaries for the authenticated agent. Policy references are masked; this does not refresh ePartner data.

list_my_ge_servicing_updates

ChatGPT
List bounded GE servicing updates for the authenticated agent. Treat stale or unknown freshness as unverified.

prepare_complete_task

ChatGPT
Prepare exactly one operation: complete the selected existing task. This preparation does not change task state. The operation is fixed by this tool name; no action selector is accepted.

prepare_move_task_to_tomorrow

ChatGPT
Prepare exactly one operation: move the selected existing task to tomorrow. This preparation does not change task state. The operation is fixed by this tool name; no action selector is accepted.

prepare_remind_task_later

ChatGPT
Prepare exactly one operation: schedule one supported reminder preset for the selected existing task. This preparation does not change task state. The operation is fixed by this tool name; no action selector is accepted.

prepare_start_task

ChatGPT
Prepare exactly one operation: start the selected existing task. This preparation does not change task state. The operation is fixed by this tool name; no action selector is accepted.

update_calendar_event

ChatGPT
Update one of the authenticated caller's personal calendar events after a clear request. Supply only the current event_id and expected_etag from the event version being edited. YCH owns trusted invocation and operation identity internally; never supply operation_id, operation_key, invocation_id, or tool-call authority fields. After UNKNOWN, refresh Calendar and do not issue a replacement identical write; YCH keeps the unresolved invocation bound to its existing operation.

Capabilities

InteractiveWrites

Example Prompts

Click any prompt to copy it.

Similar Apps