Studocu

Use this when the user wants to find, search, or browse study documents, lecture notes, summaries, or course materials on Studocu. Pass topic/keywords as `query` only — never put course name, course id, or institution name in `query`. `institution` is a display name only (this tool has no institution_id). Prefer numeric `course_id` from search-courses when you have it: it pins the course and overrules `course`. If both `course_id` and `institution` are sent, `course_id` wins; result institution is that course's university, not the name you passed. Otherwise pass `course` as name or code, never as a numeric id. `course` is optional extra if you already have `course_id`. Omit unknown filters. Returns matching documents with titles, course and institution info, ratings, and a link to view each on Studocu. Results with `match_type: semantic` are closest matches, not exact keyword hits, and may be only loosely related; check them against the request.

Search Studocu institutions by name when establishing or correcting education context (onboarding / context setup). Do not require this before every document search. Structured results include id, name, type, region_code, and country for follow-up tools. When talking to the user, present name/type/country/region only — do not show internal IDs. Auto-pick a clear top match; if the shortlist is ambiguous, ask the user to choose by name. Resolved IDs are conversation-scoped until a persistence capability owns them.

Search Studocu courses when establishing or correcting course context. Prefer institution_id from search-institutions to scope results. Omit institution_id only when the institution is unknown: results then include catalog institution_id, and institution_name when Algolia has it. Do not require this before every document search. Structured results include course_id, name, code (nullable), institution_id, and optional institution_name. When talking to the user, present name, course code, and institution name only — do not show internal IDs. Auto-pick a clear top match; if the shortlist is ambiguous, ask the user to choose by name/code. Resolved IDs are conversation-scoped until a persistence capability owns them.

Return a generated course outline as markdown when this tool has one. This is the best grounding for high-quality study guidance. If it is missing, a syllabus may be generating, so try again in a few minutes. Requires course_id.

Use this proof tool to verify that the Agent OS MCP server received a valid delegated Studocu session.

无需输入参数。

Read a Studocu project source or creation by its studocu://project/{project_id}/{kind}/{uuid} URI. Returns normalized metadata with explicitly mapped inline content.

Read a Studocu document directly, without requiring or creating a project source. Use read-artifact for documents attached to a project.

Return the unified Agent OS context manifest for the authenticated user: profile, followed courses, project metadata, and recent activity. Every project includes has_course_mastery and started_course_mastery (course supports Infinite Practice / learner has started it). Course mastery is visible only on projects, not followed_courses. Practised-topic detail (topic_masteries) is included only when a course is in scope (project_id, a sole remaining project, or remaining projects that share one course_id); the field is omitted when detail was not loaded. Empty topic_masteries means that course was loaded and the learner has no practised topics. Use get-course-mastery(course_id) for the practised-topic list when context omitted detail. Agent OS is read-only: chatting or calling this tool does not change mastery. Source of truth is Infinite Practice attempts (identity + course + topic); web IP writes, ACE/mobile/Agent OS only read.

Return sources and creations (project state) for one or more projects by id. Creation rows may include external_id (quizId or mockExamId) so a request-creation creating response can be matched once Alpha identity appears. Use after get-context (which lists projects without their sources/creations) to load those projects' artifacts. Only projects you can access are returned.

Return practised-topic Infinite Practice mastery for a course (topic_id, topic_name when available, display 0–100, questions_answered). Use after get-context flags (has_course_mastery / started_course_mastery) or whenever context omitted topic_masteries. Agent OS is read-only: chatting, creating artifacts, or calling this tool does not change mastery. Source of truth is Infinite Practice attempts (user_topic_mastery: identity + course + topic); display matches the web IP UI. Progress is learner- and course-scoped, not chat- or client-scoped — web IP writes it; ACE/mobile/Agent OS only read the same rows. Empty topic_masteries means no IP attempts yet, not a separate client score. Requires course_id.

Create a Studocu artifact from typed content. Provide project_id and an artifact envelope; the artifact type defines the required content shape.

Attach up to 10 Studocu documents from search-documents to a project. Returns the sources that were successfully added; compare with document_ids if some were omitted.

Create a Studocu project and optionally configure its emoji and course. A numeric course_id is validated before project creation.

Update a Studocu project title, emoji, or course. Use course_id null to clear the course and institution.

Generate a Studocu quiz or mockExam from 1-10 project source artifacts. Waits up to ~3 minutes; returns artifact_id and external_id when created, status creating with external_id when persist is not yet visible, or outcome_unknown on client timeout or in-flight duplicate (check get-project-state before retrying). Optional idempotency_key is forwarded as Idempotency-Key; identical retries without a key still dedupe.

Read durable agent memory for the authenticated user. Returns one document per (scope, kind); use project_id for project-scoped memory. Identity comes from Core via the session bearer, not from input.

Upsert one durable memory document per (scope, kind) for the authenticated user. Provide project_id for project-local memory, or null/omit for global. Identity comes from Core via the session bearer, not from input.

Delete one durable memory document identified by (scope, kind) for the authenticated user. Provide project_id for project-local memory, or null/omit for global. Identity comes from Core via the session bearer, not from input.