complete_playbook_step
ChatGPTRecords that the current step of a run is done and returns the NEXT step — its full instructions and optional verification — so you can carry on. Pass the run_id from start_playbook plus the step_id you just finished. Verification is optional and NEVER blocks: pass verified (true/false) and an optional verification_note to log how the step's "done when" check went, or omit them. When the last step is completed, run_complete is true and next_step is null.
create_playbook
ChatGPTSaves a process as a new, reusable playbook owned by the current user — turning something already carried out with them into a repeatable recipe. A playbook is most useful when its steps record the actual process (real commands, tool names, decisions) phrased so a later run can follow them.
get_chapter
ChatGPTReturns the teaching content, demonstration prompt, and assessment rubric for a chapter, plus any reference resources an admin attached. Gated chapters require the chapter's minimum plan (e.g. Premium, Business) or higher.
get_my_progress
ChatGPTReturns the current learner's progress across all chapters — status, start and completion dates, and score for each chapter, plus overall counts and percent complete.
get_playbook_step
ChatGPTReturns one step's instructions and optional verification check, plus its order and the playbook it belongs to.
get_team_progress
ChatGPTReturns a progress summary for members of the caller's org — each member's name, plan, completed chapter count, percent complete, and current chapter, sorted by percent complete. Results are paginated (limit/offset) so large orgs don't return the whole roster at once. Requires the manager or admin role.
get_user_progress_detail
ChatGPTReturns full progress and demonstration records for one user in the caller's org — every chapter's status and the user's demonstration attempts (most recent first) with their assessments. Demonstrations are paginated (demonstrations_limit/offset). Requires the manager or admin role; the target user must be in the caller's org.
list_playbooks
ChatGPTReturns the playbooks available to the current user — their own, system templates, and playbooks teammates shared with the org — each with its step count. Playbooks the user has hidden in the web app are omitted.
list_sales_agents
ChatGPTReturns the team's sales practice agents — saved roleplay scenarios (who the prospect is, what the rep sells, the goal, objections, script) set up on the MCP Tutor web app. Shared org-wide. One is flagged is_default: that's what start_sales_roleplay runs when no agent is named.
mark_complete
ChatGPTRecords completion of a chapter and unlocks the next one. Requires a passed demonstration for the chapter; returns an error otherwise. Returns the completed chapter and the next unlocked chapter, if any.
set_playbook_sharing
ChatGPTShares one of the user's own playbooks with their whole team (or makes it private again). Only the owner can change sharing — templates and other users' playbooks are rejected. Pass share_with_team=true to share (visibility 'org': teammates can run and duplicate it, only the owner edits) or false to make it private.
start_playbook
ChatGPTOpens a fresh run of a playbook and returns its FIRST step — instructions plus optional 'done when' verification — along with a compact outline of every step. Playbooks are stateless and repeatable: this always starts at step 1 and never resumes a previous run. Pass the returned run_id to complete_playbook_step as you work through the steps.
start_sales_roleplay
ChatGPTReturns a fully pre-built sales-call roleplay scenario — a "practice agent" the user's team configured on the MCP Tutor web app — so a practice call can open immediately, in character, with no setup questions. Covers requests like "start my sales roleplay", "let's practice a cold call", or "run the CFO agent". Unlike the 'Sales script roleplay' playbook template (which gathers the scenario interactively in chat), everything is already here; the difference matters most in voice mode. agent_id or agent_name (case-insensitive) selects a specific agent; with neither, the team's default agent is returned. The result carries the agent's fields, a markdown brief describing the prospect and the rep, and instructions for running the call and the debrief that follows it.
start_training
ChatGPTReturns the learner's active chapter — the one to work on next — plus their name, org, plan, and completed/total chapter counts. Initializes the learner's progress on first call and marks the active chapter as in progress.
submit_demonstration
ChatGPTRecords a demonstration attempt for a chapter: the learner's response, the assessment, and a pass/fail result. A failed attempt is stored but does not advance progress; the learner can retry. Returns the stored demonstration id.
whoami
ChatGPTReturns the account this connector is authenticated as — the user's name, email, role, and plan — plus the org they belong to and its member count, which is the audience any team-visible action (such as sharing a playbook) would reach. For MCP Tutor platform staff on an OAuth connector it also reports user.platform_staff: true, meaning this session can act inside other companies' orgs through the staff authoring tools.
