Name: Grade DeepTutor answers
Author: recued-core

Grade DeepTutor answers

by recued-core v7 10 views

Use “Grade DeepTutor answers” in Recued. Scheduled class batch. It uses the connections and settings you choose when you run it.

Complete workflow

DeepTutor Records 9 pinned recipes

Use DeepTutor Records in Recued for education, tutoring, student and assignment record-keeping. It stores students, assessments, answers, artifacts and scorecards, and includes 9 ready-to-run workflows that enrol students, send assessments, capture answers and grade them. It needs the DeepTutor pack for question generation. It runs locally on your machine. Actions that change data use Recued's approval controls.

You are viewing one recipe in this workflow. Installing it opens the complete pack so its other recipes and background automations arrive together.

Grade DeepTutor answers

This recipe

Use “Grade DeepTutor answers” in Recued. Scheduled class batch. It uses the connections and settings you choose when you run it.

pack:deeptutor-daily-masterydeeptutorgradingassignment

by recued-core pinned v7

Enroll free DeepTutor student

Use “Enroll free DeepTutor student” in Recued. Exact-form enrollment template for a free class.

pack:deeptutor-daily-masteryform-response-templatestudentenrollment

by recued-core pinned v4

Use “Register paid DeepTutor student” in Recued. It turns one complete Seller order into a class roster record.

pack:deeptutor-daily-masterystudentenrollmentpaid

by recued-core pinned v4

Send DeepTutor assignment or exam

Use “Send DeepTutor assignment or exam” in Recued. Scheduled class dish. It uses the connections and settings you choose when you run it.

pack:deeptutor-daily-masterydeeptutorassignmentexam

by recued-core pinned v7

Capture DeepTutor answer

Background automation

Use “Capture DeepTutor answer” in Recued. It runs on its own when a student emails their answers back, and files that reply as a pending answer so the grading workflow can score it on the next run.

pack:deeptutor-daily-masterydeeptutorstudentanswer

by recued-core pinned v5

List DeepTutor assignments and exams

Use “List DeepTutor assignments and exams” in Recued. It provides a read-only class index over the DeepTutor pack's own assessment records.

pack:deeptutor-daily-masterydeeptutorassignmentexam

by recued-core pinned v4

Show all 9 recipes

View DeepTutor assessment

Use “View DeepTutor assessment” in Recued. It provides a read-only exact detail for one DeepTutor assignment or exam.

pack:deeptutor-daily-masterydeeptutorassignmentexam

by recued-core pinned v3

View DeepTutor scorecards

Use “View DeepTutor scorecards” in Recued. Lists the derived assignment and exam scorecards the DeepTutor pack keeps in its own records. Supply one Seller customer id for that student's rows, or leave it blank for the owner-wide scorecard list.

pack:deeptutor-daily-masterydeeptutorscorecardstudent

by recued-core pinned v3

Require DeepTutor records runtime

Background automation

Use “Require DeepTutor records runtime” in Recued. It confirms this server can store the class records the other DeepTutor workflows depend on, and stops with a clear message when it cannot.

by recued-core pinned v1

View pack details, dependencies, and permissions →

Scheduled grading

Save one clone per Seller tier/class and assessment namespace. The answer batch is clamped to 1..10, but assessment question_count itself has no authored ceiling. Before any model call, each compact answer manifest drives exact file reads: Recued decodes, hashes, parses, and identity-checks both the student submission and its assessment artifact. One foreach op-step grades each verified multi-question submission as a whole.

The grader must answer exactly CORRECT|feedback, PARTIAL|feedback, or INCORRECT|feedback; those map to 100, 50, and 0. Feedback is bounded to 1,000 characters; it is stored in the answer record’s own long-text slot, not inline in a key-value envelope. Malformed output becomes needs_review with a null score and is not emailed. DeepTutor chat is not used for grading because its v1.5.5 general chat capability mounts internal tools.

Crash posture is unchanged in shape and now rides the record’s own protected revision: pending -> graded_pending_delivery before mail -> graded_delivered after mail, each update asserting the revision it read. Scorecards are recomputed only from graded_delivered answers, one scorecard record per (student, namespace). Records are pack-owned (D-221), so every row also carries the pack version it conforms to.

What the run report counts

Every batched step reports what LANDED, not what was attempted. A rejected write inside a batch is a per-item failure — the run itself still reports success — so Scorecards written counts only the writes the store accepted, Scorecard write failures counts the rest, and the scorecard table lists the rows that actually persisted. When the two disagree, the difference is real: the run did less than it set out to do.

Because this dish runs on a schedule, nobody is reading that card. Notify me when a scheduled run fails partway (on by default) sends one notification when a feedback email or a scorecard write is rejected. Nothing is lost either way — answers whose feedback failed stay at revision 1 and are retried next run, and rejected scorecards are recomputed from delivered rows next run — but without the notification the shortfall is invisible until someone opens the dish.

How it works 145 steps

Inspect the data fetches, transforms, gates, and output this recipe runs.

Process (145 steps)

■ class_key_value trim

Trim whitespace from setting class key

■ class_key_bounded_value truncate

Truncate class key value to a max length

■ class_key_bounded compare

Check if class key value equals class key bounded value

■ sender_mail default

Apply default

■ batch_limit_number to_number

Convert setting batch limit to a number

■ batch_limit_value clamp

Clamp batch limit number between 1 and 10

■ inputs_valid all

Check if all conditions are true

⚠ inputs_guard guard

Stop if inputs valid equals

■ answer_rows ?

■ pending_rows filter

Filter by condition

■ pending_count count

Count items in pending rows

■ pending_batch slice

Take a subset of

■ batch_with_submission_id map

Extract submission_artifact_ref from each item

■ batch_with_artifact_ids map

Extract assessment_artifact_ref from each item

■ submission_artifact_gets ?

■ submission_artifact_gets_ok filter

Filter by condition

■ submission_artifact_gets_failed filter

Filter by condition

■ submission_artifact_failed_count count

Count items in submission artifact gets failed

■ submission_sources map

Extract from each item

■ submission_reads ?

■ successful_submission_reads filter

Filter by condition

■ submission_reads_failed filter

Filter by condition

■ submission_read_failed_count count

Count items in submission reads failed

■ submission_texts decode_base64

Apply decode_base64

■ submission_text_rows map

Extract from each item

■ submission_hashes sha256

Apply sha256

■ submission_hash_rows map

Extract from each item

■ submission_parses json_parse

Apply json_parse

■ submission_rows map

Extract from each item

■ submission_validity all

Check if all conditions are true

■ valid_submission_envelopes filter

Filter by condition

■ valid_submission_rows map

Extract from each item

■ assessment_artifact_gets ?

■ assessment_artifact_gets_ok filter

Filter by condition

■ assessment_artifact_gets_failed filter

Filter by condition

■ assessment_artifact_failed_count count

Count items in assessment artifact gets failed

■ assessment_sources map

Extract from each item

■ assessment_reads ?

■ successful_assessment_reads filter

Filter by condition

■ assessment_reads_failed filter

Filter by condition

■ assessment_read_failed_count count

Count items in assessment reads failed

■ assessment_texts decode_base64

Apply decode_base64

■ assessment_text_rows map

Extract from each item

■ assessment_hashes sha256

Apply sha256

■ assessment_hash_rows map

Extract from each item

■ assessment_parses json_parse

Apply json_parse

■ assessment_rows map

Extract from each item

■ assessment_question_counts count

Count items in item assessment questions

■ assessment_count_rows map

Extract from each item

■ assessment_validity all

Check if all conditions are true

■ valid_assessment_envelopes filter

Filter by condition

■ grading_sources map

Extract from each item

■ grading_contexts map

Extract from each item

■ grading_requests_bounded default

Apply default

■ grading_packets map

Extract from each item

■ grade_calls ?

■ successful_grade_calls filter

Filter by condition

■ grade_calls_failed filter

Filter by condition

■ grade_call_failed_count count

Count items in grade calls failed

■ grade_rows map

Extract from each item

■ responses_trimmed trim

Trim whitespace from item raw response

■ response_parts split

Split item result into parts

■ protocol_rows map

Extract from each item

■ part_counts count

Count items in item parts

■ count_rows map

Extract from each item

■ feedback_trimmed trim

Trim whitespace from item feedback raw

■ feedback_bounded truncate

Truncate item result to a max length

■ feedback_rows map

Extract from each item

■ verdicts_trimmed trim

Trim whitespace from item verdict raw

■ verdicts_upper uppercase

Convert item result to uppercase

■ normalized_rows map

Extract from each item

■ verdict_statuses switch

Map item verdict to one of: CORRECT, PARTIAL, INCORRECT

■ status_rows map

Extract from each item

■ protocol_validity all

Check if all conditions are true

■ grade_decisions map

Extract from each item

■ decision_keys template

Generate text from a template

■ decision_rows map

Extract from each item

■ grade_scores switch

Map item decision key to one of: true:CORRECT, true:PARTIAL, true:INCORRECT

■ score_rows map

Extract from each item

■ answer_states switch

Map item decision key to one of: true:CORRECT, true:PARTIAL, true:INCORRECT

■ final_rows map

Extract from each item

■ graded_at date_add

Add 0 seconds to

■ graded_at_ms date_parse

Parse graded at as a date

■ graded_value_envelopes merge

Combine objects together

■ graded_values filter

Filter by condition

■ graded_value_sizes map

Extract result from each item

■ graded_values_within_limit filter

Filter by condition

■ graded_value_count count

Count items in graded values

■ graded_value_within_limit_count count

Count items in graded values within limit

■ graded_values_size_valid compare

Check if graded value within limit count equals graded value count

⚠ graded_values_size_guard guard

Stop if graded values size valid equals

■ grade_cas ?

■ grade_cas_success filter

Filter by condition

■ grade_cas_failed filter

Filter by condition

■ grade_failed_count count

Count items in grade cas failed

■ grade_persisted_count count

Count items in grade cas success

■ needs_review_rows filter

Filter by condition

■ needs_review_count count

Count items in needs review rows

■ delivery_rows ?

■ awaiting_delivery_rows filter

Filter by condition

■ delivery_batch slice

Take a subset of

■ delivery_batch_identified map

Extract assessment_ref from each item

■ delivery_capacity_envelopes merge

Combine objects together

■ delivery_capacity_values filter

Filter by condition

■ delivery_capacity_sizes map

Extract result from each item

■ delivery_capacity_within_limit filter

Filter by condition

■ delivery_batch_count count

Count items in delivery batch

■ delivery_capacity_within_limit_count count

Count items in delivery capacity within limit

■ delivery_capacity_valid compare

Check if delivery capacity within limit count equals delivery batch count

⚠ delivery_capacity_guard guard

Stop if delivery capacity valid equals

■ feedback_mail ?

■ feedback_mail_success filter

Filter by condition

■ feedback_mail_failed filter

Filter by condition

■ feedback_mail_failed_count count

Count items in feedback mail failed

■ delivered_value_envelopes merge

Combine objects together

■ delivered_values filter

Filter by condition

■ delivered_value_sizes map

Extract result from each item

■ delivered_values_within_limit filter

Filter by condition

■ delivered_value_count count

Count items in delivered values

■ delivered_value_within_limit_count count

Count items in delivered values within limit

■ delivered_values_size_valid compare

Check if delivered value within limit count equals delivered value count

⚠ delivered_values_size_guard guard

Stop if delivered values size valid equals

■ delivery_cas ?

■ delivery_cas_success filter

Filter by condition

■ delivery_cas_failed filter

Filter by condition

■ delivery_failed_count count

Count items in delivery cas failed

■ delivered_count count

Count items in delivery cas success

■ score_answer_rows ?

■ delivered_score_rows filter

Filter by condition

■ score_groups group_by

Group by customer_id

■ scorecard_id_seeds json_stringify

Apply json_stringify

■ scorecard_ids sha256

Apply sha256

■ scorecard_rows map

Extract from each item

■ scorecard_writes ?

■ scorecard_write_results default

Apply default

■ scorecard_write_success filter

Filter by condition

■ scorecard_write_failed filter

Filter by condition

■ scorecard_attempted_count count

Count items in scorecard rows

■ scorecard_written_count count

Count items in scorecard write success

■ scorecard_failed_count count

Count items in scorecard write failed

■ scorecard_table to_table

Format results as a data table

■ has_run_failures any

Check if any of the conditions are true

■ should_notify all

Check if all conditions are true

■ notify ?

skip: step.should_notify not_equal true

■ summary to_summary

Format results as a summary card

Settings 6 configurable

Configurable at install. Defaults shown — change them anytime in Recued.

class key setting = [object Object]

batch limit setting = 5

assessment kind setting = assignment

notify channels setting = in_app

notify on failure setting = true

sender mail instance setting =

Trust & control

What installing this recipe would let it do. Recued grants these permissions at install — review them there before approving.

Permissions it requires

Mail sendSend notifications

Declared by the recipe — Recued grants these at install, where you review them before approving.

About

recued-core

Details

145 steps 6 configurable settings recipe_id: grade-deeptutor-answers

Grade DeepTutor answers

Complete workflow

Scheduled grading

What the run report counts

Trust & control

Permissions it requires

About

Tags

Details