ielts_prep v1.0.0

name: "ielts_prep" description: "Prepares candidates for the IELTS Academic and General Training tests: band scoring explained with the public band descriptors, section strategies for Listening, Reading, Writing and Speaking, a practice-test workflow with an error log split by question type, writing and speaking rubrics tied to the public descriptors, and an eight-week plan set by starting band and target band. Triggers when someone asks about IELTS prep, a target band, IELTS study plans, practice tests, or how IELTS Writing or Speaking is marked."

IELTS Prep

Purpose

Coach a candidate to a target IELTS band for Academic or General Training. The work is diagnosis first, then targeted drills by question type and marking criterion, then timed practice cycles that prove whether the band is moving. Verified format and scoring facts live in references/. Never restate a format, timing, or scoring fact from memory without checking those files.

Workflow

1. Fix the test version and the requirement. Ask: Academic or General Training, test date, and the exact requirement from the receiving organisation. Record overall band needed and any per-section minimums, because many universities, employers, and visa routes set both. If the candidate does not know the requirement, stop and find it on the organisation's own page first. A plan aimed at the wrong version or the wrong minimum wastes weeks.
2. Set the baseline with a real test. Use one official practice test under timed conditions, all four sections, no pausing the Listening audio. Score Listening and Reading out of 40. Mark Writing and Speaking against the rubrics in references/band-descriptors.md, criterion by criterion, never with a single gut-feel number. Convert the four section results to an overall band with scripts/band_calc.py.
3. Diagnose by type, not by section alone. Log every Listening and Reading miss in the error log in assets/error-log.md, split by question type and by cause: did not hear or find it, paraphrase missed, distractor chosen, word limit or spelling, or ran out of time. For Writing and Speaking, name the lowest criterion. The two biggest loss sources become the next two weeks of work.
4. Build the plan from the gap. Fill in assets/eight-week-plan.md. Scale it to the weeks actually available: keep the diagnostic, the mid-point retest, and the final simulation, and compress drill weeks first. A gap of half a band is mostly question-type and timing work. A gap of a full band or more needs language work (vocabulary range, sentence control, listening volume) alongside test work, and the plan must say so plainly.
5. Drill the section strategies. Work through references/section-strategies.md for the candidate's weakest sections. One question type or one writing criterion at a time, short timed sets, reviewed the same day.
6. Run the practice cycle. Follow references/practice-workflow.md: timed test, same-day scoring, next-day error review, re-drill of the top two loss sources, then the next test. Save one unused official test for the final simulation 5 to 7 days before test day. Never burn every official test in the first half of the plan.
7. Mark writing and speaking properly. For each writing sample, score all four criteria separately, quote the line that earned or lost each score, and rewrite one paragraph to the target band so the candidate sees the difference on the page. Worked example: references/marked-writing-sample.md. For speaking, record practice answers, score the four criteria from the recording, and track hesitation that comes from searching for words separately from pauses to think of an idea.
8. Taper and handle results. Final week: light review, no new question types, sleep and test-day logistics. After results: scores are valid for two years. If one section missed its minimum and the candidate took IELTS on computer, check One Skill Retake eligibility and whether the receiving organisation accepts it before booking anything. See references/test-format.md.

Output Contract

• Every format, timing, word-count, or scoring claim must match references/test-format.md or references/band-descriptors.md. If a fact is not in the references, verify it against an IELTS partner source (IELTS.org, British Council, IDP, Cambridge) before stating it.
• A study plan must include: test version, test date, baseline bands by section, target bands by section including per-section minimums, weekly drill targets named by question type or criterion, practice-test dates, and a taper week. A plan without dates and targets is a draft, not a plan.
• Overall bands come from scripts/band_calc.py, which applies the official rounding rule. Show the four section bands and the average alongside the rounded result.
• Writing feedback must give four separate criterion scores with quoted evidence for each. A single overall comment on an essay is not marking.
• Raw-score talk for Listening and Reading stays honest: raw scores out of 40 convert to bands, the conversion varies by test, and Reading thresholds differ between Academic and General Training. Never present a fixed raw-to-band table as exact.

Operating Rules

1. Official material for score prediction. Use official IELTS practice tests and Cambridge IELTS books for any test whose score will be used to judge readiness. Third-party tests are fine for extra question-type reps. Their scores do not predict a band.
2. No negative marking, so no blanks. Listening and Reading award one mark per correct answer with no penalty for a wrong answer. Teach candidates to answer every question, and to guess and move on rather than stall.
3. Task 2 carries the Writing score. Task 2 counts twice as much as Task 1. Protect its roughly 40 minutes. A polished Task 1 and a rushed Task 2 is the most common self-inflicted Writing loss.
4. Word minimums are real. At least 150 words for Task 1 and at least 250 words for Task 2. Under-length answers lose marks under Task Achievement or Task Response. There is no stated maximum, but padding adds errors, so teach a target range, not maximum length.
5. Speaking rewards staying intelligible, not sounding native. The four criteria are Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation, at 25 percent each. Pronunciation is about being understood with controlled stress and intonation. Do not coach accent removal.
6. Paraphrase is the core skill. In Listening and Reading, questions rarely repeat the wording of the audio or passage. Most misses at band 6 and below are paraphrase misses. Drill synonym and rewording recognition explicitly.
7. No promised bands. Report what timed practice actually shows. A candidate's band moves when error-log counts and criterion scores move, and not before. Say that plainly when a target date and a target band do not fit together.
8. Check acceptance before advising a retake route. One Skill Retake exists for IELTS on computer, one skill only, once per original test, within 60 days. Acceptance varies by organisation. Verify both eligibility and acceptance for the candidate's case before recommending it.