assess_claim
ChatGPTRecord explicit reviewer judgment supported/partial/contradicted/unavailable. Exact excerpt is checked against retrieved text; semantic truth is NOT automatically certified.
create_research_backup
ChatGPTReturn a validated core-only ZIP as base64, not a saved download. Authorized supporting artifacts require the local CLI; no server filesystem paths accepted.
delete_my_records
ChatGPTDelete the authenticated user's retained sources, claims and document text. Other users are untouched.
delete_my_saved_project
ChatGPTPermanently remove one authenticated student's saved project, after explicit 'DELETE <project_id>' confirmation.
document_chunk
ChatGPTRead one bounded document chunk. Iterate all indices and maintain coverage before claiming a full review.
export_evidence
ChatGPTExport current user's source and claim records with candidates separated from supported evidence.
export_my_saved_project
ChatGPTReturn a verified, portable core research backup; supporting student files are not included.
format_citations
ChatGPTExecute pinned citeproc-js on metadata-checked stored sources. citations=[{items:[{id:source_id, locator:optional, direct_quote:optional}]}]. Harvard must name a supported variant. Return attribution to the student.
ingest_document
ChatGPTParse an authorized PDF/DOCX in bounded processing. Only send documents whose external processing is authorized. Returns coverage and chunk IDs, not a claim of academic review.
inspect_research_backup
ChatGPTValidate a bounded portable ZIP and reconstruct its research checkpoint without extracting to the service filesystem. Host controls actual restore/save.
list_my_research_projects
ChatGPTList only the authenticated student's saved research projects and revisions.
load_my_research_project
ChatGPTLoad signed-in student's authoritative checkpoint, evidence register, handoff, and full progress log.
openalex_lookup
ChatGPTRetrieve real OpenAlex metadata by work ID or DOI using a PyAlex-equivalent adapter.
prepare_research_handoff
ChatGPTMigrate 1.0/1.1 or validate 1.2; generate checkpoint, handoff, append-only log, evidence and manifest. Host must save returned files to authorized persistent storage; this service does not save projects.
read_passage
ChatGPTRead one retrieved passage for human/model assessment; respect quotation limits.
record_research_activity
ChatGPTAppend substantive activity/correction and regenerate consistent files. Explicit student confirmation required for approval changes. Check host's current revision before replacing saved files.
research_check_edit
ChatGPTSupport Academic Humaniser edits: flag changed numbers, quotations, citation markers and qualifications. No authorship detector.
research_check_integrity
ChatGPTCompare a chapter with visible owned-project evidence using Noplag and AEGIS. No web-wide plagiarism score.
research_check_references
ChatGPTUse Mark Russinovich RefChecker on structured references, with explicit consent for Crossref metadata queries.
research_compare_citation_patterns
ChatGPTCompare caller-resolved reference identities; repeated citation order is not proof of copying.
research_resolve_review_issue
ChatGPTRecord an explicit accept/reject/defer decision; does not rewrite a chapter or approve methodology.
research_review_context
ChatGPTRead owned blueprint, approvals, evidence and unresolved gaps for chapter planning/review. No automatic supervisor judgement.
research_save_review
ChatGPTSave proposed chapter findings to the original versioned project; preserve all approvals.
research_search
ChatGPTSearch crossref, openalex or semantic_scholar. Returns candidates, never claim verification.
resume_project
ChatGPTValidate/migrate portable checkpoint and detect missing files or decision conflicts; never imply automatic cross-chat persistence.
resume_research
ChatGPTReconstruct authoritative JSON from both generated Markdown files, reconcile attachments/evidence and approved decisions, return concise progress and essential blockers. No live citation re-verification or cross-chat memory.
retrieve_passages
ChatGPTFetch permitted PLOS open-access section text or label an abstract-only/unavailable fallback. Does not decide semantic support.
save_research_project
ChatGPTSave a versioned research project durably for the signed-in student only. Explicit storage consent required; supporting attachment bytes are not retained.
service_status
ChatGPTReport executed service components, versions, limits and license attribution.
update_my_research_project
ChatGPTAppend research activity with durable optimistic revision check. Approved decisions need explicit student confirmation.
verify_source
ChatGPTCheck authoritative metadata and Crossref notices; save to current user's register. Supply expected title/authors/year/venue to detect mismatches.
workspace_m5_document_status
ChatGPTRead the processing status of a submitted document job in the signed-in user's owned project. Requires project_id and job_id. Does not change stored documents or certify citation accuracy.
workspace_m5_find_evidence
ChatGPTSearch indexed, non-withdrawn evidence belonging to the signed-in user's project. Return matching passages with provenance. Requires an owned project_id and query; retrieved matches do not establish claim support or plagiarism.
workspace_m5_submit
ChatGPTStore an authorised PDF, DOCX or text document in the signed-in user's owned project and queue extraction and indexing. Requires explicit consent_to_store and the current notice_version from service_status. Returns a job identifier; submission does not mean indexing or citation verification succeeded.
workspace_m5_withdraw
ChatGPTWithdraw a document from the signed-in user's owned project: hide it from evidence retrieval and queue database revocation. Requires project_id and document_id and an explicit user request. This does not certify deletion from backups.