Instructors using Schoology or Canvas can route agent-drafted rubrics, feedback comments, and grade suggestions through Dock before they touch a student record. Each agent action becomes a Dock row with a link back to the Schoology assignment, the agent identity that wrote it, the instructor who signed off, and the timestamp. The LMS still grades. Dock holds the interpretation layer so review is fast and attribution is intact.
Schoology and Canvas stay the system of record for the raw data. Dock is the system of record for what the AGENT INTERPRETS. Each Dock row carries a pointer back to the platform record, agent identity, decision, reviewer, and timestamp. The agent re-fetches platform data via fresh API reads when it needs current state.
The instructor review surface
| Assignment | Agent draft | Agent ID | Instructor | Decision | Schoology link |
|---|---|---|---|---|---|
| ENG-204 Essay 3 | Rubric brief, 4 criteria, 1-4 scale | mei-rubric-v2 | M. Chen | Approved | sch.ly/a/8821 |
| BIO-110 Lab Report 2 | Feedback comments on 18 submissions | flint-feedback-v1 | R. Okafor | Approved w/ edits | sch.ly/a/9104 |
| HIST-301 Source Analysis | Grade suggestions, 22 submissions | mei-grader-v3 | A. Patel | Sent back, bias flag | sch.ly/a/9233 |
Every row carries a Schoology assignment ID, the agent that drafted it, the reviewing instructor, and the disposition. Bias flags, edits, and rejections live on the same row, so an end-of-term audit pulls a complete trail. The pattern matches what we describe in agent audit and compliance.
A worked workflow
A composition instructor opens essay 3 in Schoology and pings the rubric agent. The agent re-reads the assignment prompt, the syllabus learning outcomes, and three exemplar essays through the Schoology API. It writes a four-criterion rubric brief and posts it as a Dock row tagged to the assignment ID. The instructor opens Dock, reviews the brief in two minutes, edits one criterion description, and approves. Dock writes the approved rubric back to the Schoology assignment. The agent then drafts feedback comments for each of the 31 submissions, posts them as a second Dock row, and waits. The instructor accepts 27 comments verbatim, edits 3, rejects 1. The rejection sits on its own row with the instructor's reason, which feeds the next round of agent tuning. The same review pattern works for any agent action that touches student records, which is why we use it across Dock for education and Dock for research.
Why it matters
Schoology Learning supports about 7 million students and delivers roughly 186 million assignments per year according to PowerSchool's product page. Canvas reports tens of millions of users and a thousand-plus third-party integrations on the Instructure overview. Agents are already touching that volume of coursework, often through ad-hoc browser extensions or copy-paste. Without an interpretation log, an instructor cannot reconstruct who suggested what, which means accreditation reviewers and student appeals land on a guess. Dock fixes that by carrying agent identity and decision history at the row level, the same approach we take for design review in Dock for design.
Try the Dock education rail with one course and one agent. Start with rubric review, then move to feedback, then grade suggestions. The audit log compounds from day one, which is what audit and compliance reviewers actually ask for.
FAQ
Does Dock replace the Schoology gradebook? No. Schoology and Canvas remain the gradebook of record. Dock holds the agent draft, the instructor decision, and the link back.
What if the agent reads stale assignment data? The agent re-fetches from Schoology on each run. Dock stores the interpretation, not a cached copy of the student work.
Can two agents share the same Dock surface? Yes. Each row records the agent identity that drafted it, so a rubric agent and a feedback agent can write to the same assignment surface without collisions.
How does this support accreditation review? Every agent action is one row with assignment pointer, agent, reviewer, decision, and timestamp. Filter by term and export.
