Competency Transcript Generation

A competency transcript is a narrative companion to the traditional grade transcript. It explains which competencies the student has demonstrated and which mastery-map cells and assessments support each claim. It does not replace credits, GPA, or conferral. Student information systems such as Ellucian, Workday, and Anthology remain the official academic record. The learning management system, often Canvas, remains the store of assignment artifacts, rubric rows, and timestamps. The language model drafts prose from those two layers. It does not become a second registrar.

Treat every generated paragraph as proposed language until the registrar issues the document. If the narrative claims mastery and the SIS still shows an incomplete or a failing mark in the only course mapped to that competency, stop. Fix the academic record through the usual process. The narrative cannot cover a missing credit or rewrite a posted grade.

Degree progress and conferral stay on the audit path. Use automated degree audit to decide whether requirements are met. Use this workflow only to decide whether the competency story is evidence-backed and fit to sit next to the grade transcript.

Generation starts with a data pull, not with a prompt. Export the current mastery map for the student: competency identifiers and labels, mapped courses or modules, cell states (not started, in progress, demonstrated, not yet demonstrated), and the last assessment that moved each cell. Export the matching assessment file: instrument name, date, rater or auto-score source, rubric criterion if one exists, and the recorded score or performance band.

Join the two files on competency ID. A cell with no matching assessment row is unsupported. A cell that moved on a quiz, a performance task, and a clinical observation should carry all three citations when all three rows are present. Do not collapse them into a single mention of coursework.

Prior learning belongs in the draft only after it is on the official record or in the PLA file the registrar already accepts. Do not fold informal work history into transcript language. Run that evidence through prior learning assessment evaluation, then cite only what that process accepted.

If cell states were inferred rather than entered by an assessor, read competency mastery inference before you generate prose. Inferred cells need a stricter review than cells an instructor marked.

Illustrative example: a nursing program maps sterile technique to a skills-lab checklist, a simulation debrief rubric, and a precepted clinical observation. The map shows demonstrated on the checklist and the simulation, and empty on the clinical observation because the rotation has not started. The draft should name the two completed instruments, their dates, and their cell IDs, and leave the clinical line blank. It should not claim the student is prepared for sterile procedures in clinical settings on lab evidence alone.

Each competency paragraph should open with the claim the map actually supports, then list the evidence. Name the competency ID and label. Repeat the cell status in the same words the map uses. Cite the assessment title, date, and result. If a rubric row is what moved the cell, cite that row, not only the assignment title.

Do not translate a B-minus into strong competency or a pass into mastery. The grade transcript already reports the grade. The narrative reports demonstrated competencies and the instruments behind them. Mixing the two is how a reader starts treating the narrative as a grade change.

Keep the scope of each sentence inside the evidence. A usable line names the week-6 skills checklist, cell ST-04, and the simulation rubric row that covers maintaining a sterile field, with the dates those instruments were scored. An unusable line declares mastery of sterile technique while the map still has an empty clinical cell for that competency.

SIS suites such as Ellucian, Workday, and Anthology, and an LMS such as Canvas, export competency frameworks and gradebooks in different shapes. Do not assume a product-specific competency engine. Require a stable competency ID, a cell state, and an assessment pointer you can reprint on the transcript. If the export cannot produce those three fields, stop generation and fix the data contract.

When the issued narrative later travels as a badge or verifiable credential, package only what the registrar signed. See digital credential packaging. A polished paragraph that never passed review is not packageable evidence.

Empty stays empty. If a competency has no assessment row and no demonstrated cell, the transcript line is blank. Do not write a hedge, a developing story, or a sentence about related coursework under a different competency ID. Filling a blank competency invents mastery, even when the rest of the map looks complete.

Thin is not the same as empty. Thin means some evidence exists but not enough for the program's issuance rule, for example lab only when the map requires lab plus clinical. For thin cells, either omit the competency from the issued narrative or print the cell status and the missing instrument without promoting the status to demonstrated. Do not invent the missing observation.

Do not let the model finish a paragraph because neighboring competencies in the same domain are demonstrated. Adjacent evidence does not transfer. The map is a grid of cells, not a general impression of the student.

Do not backfill from course grades. An A in the lecture does not fill a blank competency line unless a mapped assessment moved that cell. Leave the line blank and send the gap to the assessment lead. Do not send it back to the model with instructions to say something anyway.

The registrar owns what is official. Assessment leads own whether the cited instruments are the right ones and whether cell states match the map. Skip neither role.

Review the draft against the two source files, not against how the prose sounds. Check every citation: competency ID, cell status, assessment name, date, result. Check every blank: confirm the source files are also blank, so the model did not drop evidence, and confirm the model did not add a sentence where the files are empty. Check that no paragraph asserts a grade, a GPA effect, or a conferral the SIS does not show.

Do not publish an unreviewed narrative. A PDF in the student portal, a paragraph pasted into a recommendation letter, or a file sent to an employer is an issued claim as soon as it leaves the review queue. Publishing unreviewed narrative puts unofficial language next to an official transcript.

After approval, freeze the issued text with the source extract used to generate it, meaning the map snapshot and the assessment file, each with a timestamp. Later map updates must not silently rewrite an already issued competency transcript. If mastery changes, issue a dated revision through the same review gate.

If a reviewer believes the narrative exposes an error in the official record, open the petition or change-of-grade process. Do not treat the narrative as a grade change. Do not let a student argue that a well-written competency paragraph overrides a posted mark.