august

Turn a one-line research brief into a validated survey question set — the questions array that create_survey accepts — without hand-authoring questions. August's survey-design intelligence picks the question types (poll / maxdiff / rating / open_text), writes balanced prompts and options, and validates the design against the same rules create_survey enforces. Returns {title, research_goal, participant_count, segment_ids, questions, ready, assistant_text, missing_fields, validation_warnings} — the questions, research_goal, title, participant_count, and segment_ids are already in create_survey's shape, so pass them straight through (check `ready` first). Prefer this over writing questions yourself when the user gives a goal rather than specific questions: call design_survey, then create_survey, then run_survey. Makes one metered LLM call.

design_survey

End a running focus group, generate its report, and return a compact completion confirmation. Fetch report contents separately with get_focus_group_results.

end_focus_group

Fetch the latest report for a completed focus group: summary, conclusions, per-round outputs, insights, and dissenting views. Use after a focus group finishes running in the August app.

get_focus_group_results

Fetch the full transcript of a 1:1 twin conversation (interview) by conversation id: conversation metadata plus every message with its role and content. Long messages and very long conversations are truncated.

get_interview_transcript

Fetch one twin segment by id: its definition, build-run status, and matched twin count.

get_segment

Run the guided first August interview: resolves the user's organization, picks a ready twin, asks it one short starter question for the organization's vertical, and returns the twin's answer with suggested follow-ups. Prefer this for a user's first session with August, or whenever they ask how to get started or what August can do. If the user has no organization yet, the result explains exactly what to collect before calling create_organization; after the organization is created, offer this tool as the example first interview and call it when the user accepts. Once an interview is open, continue with ask_question(interview_id) rather than calling this again.

getting_started_with_august

Fetch one survey by id: its questions, participant roster, status, and — once the run has completed — its report. For just the aggregated results use get_survey_results; for a running survey's live counts use get_survey_progress.

get_survey

Fetch a completed survey's report (aggregated per-question results and analysis). Set compact=true for a bounded summary suitable for Slack. When no report exists yet this returns {status:'not_ready', poll_after_seconds} instead of erroring. Check get_survey_progress: if status is 'ready' the survey was never started — call run_survey (do not poll); while 'running', keep polling and retry once 'completed'; if 'failed' or 'cancelled', the survey produced no report — stop.

get_survey_results

List recent 1:1 twin conversations (interviews) the user can access, with twin name, title, status, last-message preview, and message count. Pass org_id whenever known (from list_organizations or create_organization); omitting it only works when the caller's token carries an organization, and sessions that created their organization here do not.

list_conversations

List focus groups (multi-twin research sessions) the user can access, with status (draft, config, ready, running, complete, archived) and metadata. Pass org_id whenever known (from list_organizations or create_organization); omitting it only works when the caller's token carries an organization, and sessions that created their organization here do not.

list_focus_groups

List the twins that belong to a segment (from its most recent successful build), with each member's match score and short rationale. Returns an empty list if the segment has not been built yet. Paginated: the response includes a nextCursor when more members remain — pass it back as cursor to fetch the next page.

get_segment_members

List the saved twin segments (named audiences) in an August organization, with each segment's id, name, latest build-run status, and matched twin count. Pass a segment's id as segment_ids on create_survey to poll that audience instead of the org's newest twins. Paginated: pass a returned nextCursor back as cursor to fetch the next page.

list_segments

List the surveys in an August organization, each with its status, title, participant count, and response counts. Use a survey's id with get_survey / get_survey_progress / get_survey_results. Paginated: pass a returned nextCursor back as cursor to fetch the next page.

list_surveys

List the ready digital twins (personas) in an August organization, returning up to the requested limit (100 by default) with search-card profile fields and avatar. Use a twin's id with conversation tools to interview them.

list_twins

List the August organizations the authenticated user belongs to. Call this first to discover org_id values for the other tools. When the user has no organization yet, the result explains how to set one up (create_organization, or accepting a pending invitation).

list_organizations

Run one focus group discussion round and wait for participant responses, moderator synthesis, key insights, and consensus. Ask one clear question at a time and pass the user's wording faithfully.

run_focus_group_round

Start (execute) a survey that is in 'ready' status. This is the metered step: it gates on credits and returns immediately (HTTP 202) — the run itself is a long-running background job, so this does NOT return results. It returns {survey_id, status, progress}, where status is normally 'running' (or 'completed' if the survey had already finished). Then poll get_survey_progress until status is 'completed' and fetch get_survey_results.

run_survey

Select the ready twins best suited to a research request. The selection uses the shared profile/segment reranker used by research panels, then returns the selected twins with their profile summaries.

select_twins

Begin a ready August focus group and return its live state and participants. Use this when the user wants to start or begin a specific focus group. After starting, run discussion rounds with run_focus_group_round.

start_focus_group

Begin a 1:1 interview with a digital twin and return an interview_id. Use this when the user wants to interview/talk to a specific twin. Then relay the user's questions to the twin with ask_question(interview_id). This only opens the session — it does not ask anything yet.

start_interview

Legacy synchronous single-twin Figma review. Prefer start_figma_review plus polling get_interview_transcript for normal design reviews because a detailed twin response can outlive the MCP request timeout. This tool asks one twin and waits for the answer. August renders the supplied Figma URL and attaches a compact layer tree before the twin responds. Use this instead of ask_question for visual Figma feedback. Check get_figma_connection first; the tool returns a clear connection action if Figma has not been connected.

Read the exact selected Figma node and return a bounded text summary of its visible structure and copy for choosing relevant twins. This selection-only tool does not render an image and must not replace start_figma_review for the actual visual review.

Start an asynchronous Figma review in 1-3 existing twin interviews. Pass every selected interview id in one call. Each conversation owns its background task, and the tool returns immediately with a task_id plus the same interview_ids. Poll each conversation with get_interview_transcript(conversation_id) until figmaReviewTask.status is succeeded or failed. Use this lifecycle for normal Figma reviews so long twin responses do not depend on the MCP request timeout.