Study Sprint Evidence Dossier
Use this exact fictional record when prior project evidence is unavailable. It is an equal-credit evidence route, not a model answer.
Evidence status: supplied course record, dated
2026-08-31. Say “the supplied record reports,” not “I observed,” unless
you separately ran an approved copy. No real student, private data,
outcome study, or live AI system is represented.
Bounded claimOnly what the named evidence supports.
Evidence routeWhere the record came from.
DeterministicThe same fixed rules handle the same input.
Artifact boundaryWhat is included and excluded.
Human-control recordThe decision or approval that stays with a
person.
Record A. Demonstration
- User and task
- A fictional learner wants to divide one study task into a short session.
- Input
-
Task:
Review chapter 4. Duration:25 minutes. - Supplied output
- Prepare: Open chapter 4 and choose two headings. Focus: Read one heading and write two notes. Reflect: Write one question for the next study session.
- Bounded claim
- The supplied record demonstrates a three-step planning interaction.
- Not established
- Learning, grades, time on task, usefulness for every learner, reliability, or production readiness.
Record B. Human process decision
- Initial proposal
- Use AI to personalize the plan and describe it as improving study outcomes.
- Human decision
- Reject the outcome claim and live-model dependency. Keep fixed Prepare, Focus, and Reflect rules.
- Reason
- The course record can show the interaction, but it contains no outcome evidence.
- Result
- A deterministic static prototype with plain-text output and no required account or model call.
- Tradeoff
- The result is easier to inspect but does not provide adaptive personalization.
Record C. Source trail
- Source identity
- Course-owned Study Sprint Behavior Specification, this dossier, Records A and B, 2026-08-31.
- Exact support
- “For a fictional task and duration, return three labeled steps: Prepare, Focus, and Reflect. Do not claim an effect on learning or grades.”
- Claim status
- Source-supported supplied record.
- Additional check
- Compare the claimed labels and limits with Record A before presenting.
- Limitation
- This is a course-owned synthetic specification, not independent evidence of real-world effectiveness.
Record D. Supplied rendering retest
Expected
Review <b>chapter 4</b> appears
literally.
Original observationThe browser interpreted the tags, so
the tag characters were not visible.
Minimum fixPlace learner-provided text with a plain-text
rendering path rather than interpreting markup.
Supplied retest
Review <b>chapter 4</b> appeared
literally. Ordinary text still displayed.
Bounded finding: the supplied record reports literal display for the named retest path after the fix. It does not prove complete security or correct behavior for every input.
Record E. Assistance, feedback, limitation, and next evidence move
Assistance disclosure
AI helped propose wording and a dynamic personalization idea. A human rejected unsupported outcome claims, kept fixed rules, chose plain-text rendering, and checked the supplied records. Final responsibility remains human.
AI helped propose wording and a dynamic personalization idea. A human rejected unsupported outcome claims, kept fixed rules, chose plain-text rendering, and checked the supplied records. Final responsibility remains human.
Fixed reviewer comment
Notice: the evidence chain is visible. Question: what supports offline behavior? Next test: open an approved copy with networking disabled and record the result.
Notice: the evidence chain is visible. Question: what supports offline behavior? Next test: open an approved copy with networking disabled and record the result.
Calibrated response
Decision: Test. Revision: do not imply offline operation. State that offline behavior is unknown until the named test is run. Stop if the route would expose private or restricted data.
Decision: Test. Revision: do not imply offline operation. State that offline behavior is unknown until the named test is run. Stop if the route would expose private or restricted data.
Known limits: no real-user, learning-outcome, broad accessibility, offline, storage, reliability, or complete-security evidence is supplied. A peer is optional: a student may use the fixed comment above, structured self-review, or instructor review for equivalent feedback evidence.