Laya UPSC

After confirmation, move an unfinished entry or place a pin on a product day.

move_plan_entry

The reading card's own call: short-lived sealed access to the figures of the page it shows, for the signed-in reader.

open_course_figures

The user's graded attempts, newest first: attempt id, item, score, band, pass, source, when. Page with before.

list_attempts

After confirmation, pin an owned thing for a Monday product week. Placement or apply chooses the date.

pin_plan_thing

Read whether graded practice is available to the connected reader, the remaining practice capacity as a fraction, and its reset times. Check before asking a practice question and again before submit_attempt. If unavailable or exhausted, explain the restriction neutrally and offer ungraded recall or saving an answer. This read does not reserve capacity, grade an answer, change account access or initiate any purchase.

get_budget

What Laya says the user should practise, per lane, track 'gs' (default), 'optional' (their active Optional), or 'all': due bricks and items, ripe questions each with a reason_string, intake, and active_optional {root, title}, null when no Optional is chosen. Laya's own questions are the GS bank and, for some Optional subjects, their past papers, and those surface only in the user's active Optional. Narrate it as given, reason strings VERBATIM; never compute, reorder, or invent recommendations. Ask a ripe question verbatim with its paper and marks.

get_practice_queue

Laya computes the schedule and returns effective preferences, product dates, reasons, left-outs and bank-add count. Two-call default: propose once, show the preview and ask for a plain yes, then apply once. Never list the shelf or compute the schedule yourself.

propose_plan

Read one page of a Foundation Course part, about 600 words, with figures as marker lines. Pass page for a page number from 0, or resume true for the page where the reader stopped. Show the text as written. The reading card shows the page and its figures where ChatGPT supports it. After the reader has read a page, call save_reading_progress with its end_percent, which needs no confirmation. Signed out, each topic's opening part alone can be read.

read_course_part

Read one search result in full by its id: a Laya question with Laya's model answer and value addition, a Foundation Course part, a Mains GS Thread with its questions in year order, or one of the reader's own Mains Notes answers with its bricks and revision card. Cite the url it returns.

fetch

One graded attempt in full, the transcript, the scorecard, every brick verdict with its text. Quote it; never restate a score the payload does not carry.

get_attempt

Read your activity for one IST day within the past year. Returns counts and at most20 entry summaries, with the full count and a record link. Do not invent activity for omitted entries.

get_activity_day

One of the user's Mains Notes items in full: question, answer, value addition, revision card, mnemonics, paper/year/marks, node paths, every brick with its id, text and starred flag, and edited (when each facet last changed). Read before any edit (update_copy_item replaces the whole item) and for informal quizzing from the user's own bricks in ungraded mode. Quote the item's own text; never add what it does not hold.

get_copy_item

Read bounded product days, titles, minutes, completion and deep links.

get_plan

Read the reader's Laya plugin settings: what a save echo shows, the marks to assume when none are named, value addition depth, whether to offer recording an evaluation, automatic planning, the default planning days, and two study plan switches. Read once in a conversation before drafting an answer to save.

get_settings

Read planning time, balance, subjects and targets.

get_plan_prefs

Read your own sittings in the last7,30 or90 IST days. Returns window totals plus a page of start dates, minutes, targets and status. No Baithak member details or tag names are returned.

get_riyaaz_log

Read a page of your saved outside evaluations: answer title, marks, maximum, verdict, date and answer link. These are saved reports, not Laya-graded attempts or revision ratings.

list_evaluations

Read a page of visible syllabus nodes with your strong, weak, due and untouched counts and weak fractions. Ordered by due then weak counts; use total and offset for more. Never returns memory parameters.

get_map

Read a page of your sets, with titles, dates and visible answer counts. This read changes nothing. Use the set link to inspect its answers; do not assume an empty count means the set was deleted.

get_sets

One bounded private study summary: streak, Riyaaz rhythm, activity, badges, coverage, visible mastery states and weak bricks, Optional, Dohraan and Plan weeks, sets, evaluations, recent checks, AI fractions and exam dates. Call once for how am I doing or what to focus on. Use exact payload numbers; omitted detail is not zero. No memory parameters, money or member identities. Limited to 30 calls/hour inside the read rail.

get_insights

Read your activity totals for the requested window, current streak and the badge substrate with explicitly named lifetime and trailing windows. Badge windows keep their own periods; do not relabel them as the requested window.

get_record

Read up to10 of your weakest visible bricks, with text, weak state, due flag and answer links. Text and titles are excerpts of at most160 characters; Memory numbers never leave Laya. This is analysis, not a revision schedule.

get_weak_bricks

Record outside feedback without Laya grading or revision-memory changes. Confirm selected questions and supplied-solutions vs independent judgement. Keep context: basis, test/question IDs, scope, actual sources/pages and limitations. Echo context, source, scores/unscored, full report and destination; wait for a plain yes before confirmed:true. Use item_id or exact question: one match appends, none creates, ambiguity refuses. Disclose differing existing answers; preserve originals. Reuse idempotency_key only for identical retries. Use import_copy for multi-answer batches.

record_evaluation

Record explicitly confirmed self-check ratings from get_revision_session with the app's own scheduling. Echo the actual brick texts and1 missed/2 hard/3 recalled; wait for a plain yes. Send item_result null and the fixed source value, never an outside evaluation score or invented recall. Up to50 events atomically; free accounts share five distinct questions per product day across app and connector. Use fresh event UUIDs and current rated_at; identical retries reuse both. No paid model call.

submit_review

After confirmation, remove the Plan entry. Its underlying thing is kept.

remove_plan_entry

Read up to50 own answers in requested or set order, with revision bricks, skipped reasons and the same free-five allowance as the app. Supply item_ids OR set_id; use next_after for the next set page. Never invent missing bricks or ratings. Offer one question at a time and ask the user to self-check. This read changes no memory and reserves no allowance.

get_revision_session

Replace one answer's mnemonics, the user's own memory aids (up to 10; each up to about 100 words, with an optional short label saying what it is for). Use ONLY when the user asks to keep or change a mnemonic; craft one for them on request using your own reasoning (Laya spends no AI on this). Read the item first (get_copy_item), send the WHOLE list, existing entries with their id plus the new one, and never touch bricks or the answer for this: it is not an edit, nothing is regenerated.

set_mnemonics

Save how far the reader has read in a Foundation Course part, as a percentage, usually a page's end_percent. Call it after each page the reader has read, without asking: progress only moves forward and stops at 99. Set completed true only after the reader says yes to marking the part done. That sets 100 and ticks the part in Laya and in the reader's plan.

save_reading_progress

Save ONE confirmed question with its answer, bricks and any value addition or revision card to the user's Mains Notes. Before calling: star the must-hit bricks (at least one, at most a third), tag it (get_syllabus, the paper's tree or the Optional it marks active and never another, the deepest nodes that fit, in node_paths), then confirm-echo the question, every brick with its stars, the answer's first line and the node titles, and get a plain yes. Response mode 'saved' = live in their Mains Notes (item_id, item link); 'inbox' = it went to the Import inbox for review. Never say saved when the response did not. Multi-question material goes through import_copy, never single saves.

save_copy_item

Search Laya's published UPSC Mains questions with Laya's model answers, Foundation Course notes and Mains GS Threads, and, when the reader is signed in, their own Mains Notes. Returns up to ten results, each with an id, a title and a getlaya.in url to cite. Use fetch with a result's id to read it in full.

search

Full-text search across the user's Mains Notes and Laya's own questions. source 'mine' = the user's items only, 'laya' = Laya's questions only, 'all' (default) = both. Laya's rows are the GS bank plus the past papers of the user's active Optional; the user's own rows can sit under any subject. Results carry ids for get_copy_item / update_copy_item / move_answer_to_trash; distinguish Laya's items from the user's when you narrate.

search_copy

After confirmation, update only the named preferences.

set_plan_prefs

Submit the user's spoken or typed answer to a Laya question as a real graded attempt, same grader, same budget and quotas, same memory writes as the app; source stamped with this platform. Probe get_budget first and again right before submitting; show the transcript back and let the user correct it before this call; never call while a budget or free-week wall stands. Returns the scorecard: score, band, pass, off_topic, each brick by text as evidenced/partial/absent with its evidence, structure, practice_only, budget_remaining, and the deep_link.

submit_attempt

The public UPSC syllabus Laya tags by. Omit root: the roots {root, kind, title, active} and active_optional {root, title} or null, gs1–gs4 (GS1 heritage/history/geography/society, GS2 polity/governance/social justice/IR, GS3 economy/S&T/environment/security, GS4 ethics) and every Optional; the root with active true is the only Optional you may tag under. Pass one root: its nodes {path, title}, GS is paper › topic › sub-topic (the topic itself is a legal tag), an Optional paper › unit › topic. Tag every capture under ONE root: the deepest node(s) that fit, best fit FIRST (the primary decides where the item lives), confirmed by the user as a path of titles; untagged only when nothing fits.

get_syllabus

After confirmation, change only the done mark. No review or paid check is submitted.

tick_plan_entry

After confirmation, undo within 24 hours. Keeps completed work and bank Notes; refuses conflicting edits.

undo_plan