✎ Student AI Disclosure Kinetic Gain Protocol Suite
EdTech trio · student-side

What did the student
actually use AI for?

Student AI Disclosure is an open JSON spec attached to student-submitted work: AI tools used (with back-refs to Agent / Tutor Cards), 13-value role taxonomy, assistance extent, prompt evidence (full / hashed / omitted), artifact-hash binding, AUP reference, signature + teacher acknowledgment. The student-side of the EdTech trio.

Read the spec Try the visualizer
  • Detect via disclosure_version
  • Well-known: — (travels with the artifact)
  • Cross-referenced from AI Incident Cards on failure

Why disclosure is the lawful state

Most academic-integrity policies treat any AI use as suspicious. This spec inverts the framing: disclosure is the lawful state. A disclosure with ai_used: false is valid. A disclosure with ai_used: true and a list of roles is valid. Missing disclosure on an assignment that requires one is the failure mode — not AI use. The spec punishes omission, not use.

AI usage facts

ai_used bool, tools_used[] with agent_card_uri / tutor_card_uri back-refs, roles[] from a 13-value taxonomy, assistance_extent (minor / substantial / primary_author), optional assistance_pct.

Prompt evidence — three modes

full (literal text preserved), hashed (canonical SHA-256 only — privacy-preserving), omitted (consciously not retained). Schema gates which fields can appear per mode.

Artifact binding

artifact_hash = canonical SHA-256 over the submission bytes. Binds the disclosure to a specific file. Post-submission edits become detectable in seconds.

AUP reference

aup_uri points at the operative Classroom AI AUP at submission time. A grader's compliance check is a single JSON join.

Signature + acknowledgment

signed_by_student: true + student_signature_at required. Optional teacher_acknowledged block (by + at + note) captures grader review.

Closes the EdTech trio

Joins with Tutor Cards (vendor-side) and Classroom AI AUP (district-side). Three documents, two joins, one allow/deny answer per submission.

The required sections

  1. disclosure_version — must be "0.1"
  2. student — pseudonymous id, optional display_name, grade_or_year, institution_id
  3. assignment — id, title, course_id, LMS name, due_at
  4. ai_used — boolean gating all AI-related fields
  5. tools_used / roles / assistance_extent / prompt_evidence_mode — required when ai_used is true
  6. artifact_hash — SHA-256 binding to submitted bytes
  7. aup_uri + policy_compliant (declared) — AUP join
  8. signed_by_student + student_signature_at — required
  9. teacher_acknowledged (optional) — by + at + note

A canonical example

{
  "disclosure_version": "0.1",
  "disclosure_id": "d-2026-05-12-b4f9c1e8",
  "created_at": "2026-05-12T16:42:00Z",
  "student": { "id": "stu-9c2e44", "grade_or_year": "11" },
  "assignment": {
    "id": "assn-2026-bio-lab-7",
    "title": "Cell Respiration Lab Report",
    "course_id": "course-bio-11-spring-2026",
    "lms": "canvas"
  },
  "ai_used": true,
  "tools_used": [
    { "name": "Claude.ai", "provider": "Anthropic", "version": "claude-sonnet-4-6" }
  ],
  "roles": ["edit", "cite_check"],
  "assistance_extent": "minor",
  "assistance_pct": 8,
  "prompt_evidence_mode": "hashed",
  "prompts": [
    { "id": "p1", "hash": "sha256:c7d1f2a3b4c5d6e7f8a9b0c1d2e3f4a5b6c7d8e9f0a1b2c3d4e5f6a7b8c9d0e1",
      "at": "2026-05-12T15:55:00Z" }
  ],
  "artifact_hash": "sha256:1a2b3c4d5e6f7a8b9c0d1e2f3a4b5c6d7e8f9a0b1c2d3e4f5a6b7c8d9e0f1a2b",
  "aup_uri": "https://lincoln-high-district-42.edu/.well-known/ai-aup.json",
  "policy_compliant": { "declared": true,
    "reason": "AUP §3.2 permits AI-assisted grammar review and citation verification for lab reports." },
  "signed_by_student": true,
  "student_signature_at": "2026-05-12T16:42:00Z"
}

Tooling around the spec

Specification

student-ai-disclosure-spec

Normative spec, JSON Schema 2020-12, canonical examples. AGPL-3.0 for spec text; implementations unrestricted.

View repo →
Visualizer

kinetic-gain-visualizer

Unified visualizer for all 10 specs. Auto-detects via disclosure_version and renders a procurement-grade view.

Open visualizer →
MCP server

mcp-kinetic-gain

34 tools across 8 specs. Drops into Claude Desktop / Cursor / any MCP-compatible client via stdio with one config entry.

View on GitHub →

About the Kinetic Gain Protocol Suite

Student AI Disclosure is one of ten open JSON specifications in the Kinetic Gain Protocol Suite. Five core specs plus the EdTech trio, the HealthTech extension, and the cross-cutting Incident Card. Front door: suite.kineticgain.com.