| name | honor-disclose |
| description | Generate an honest, structured disclosure block for a KTH submission — what was the student's own work, what help was used (people, sources, AI), and where each contribution appears. |
honor-disclose
Produce a disclosure statement that satisfies Rule 2 (disclosure of help and sources) and Rule 7 (generative AI) of the KTH EECS Code of Honour.
The principle: it is not the receipt of help that violates the code — it is the failure to disclose it. This skill makes disclosure routine.
When to use
- Before submitting an assignment, lab report, project, or thesis chapter.
- Whenever a session has involved AI assistance, peer discussion, or use of external code/text/figures.
Inputs to collect
Ask the user (in one consolidated turn — don't drip-feed questions):
- Submission identity — course code, assignment name, author(s).
- Permitted-aids policy — what the assignment explicitly allows (paste it, or point to it). If unknown, say so; the disclosure will note this.
- Help received, broken down:
- People — who, in what role (peer discussion, TA hint, supervisor feedback, proofreader). Names if applicable.
- Sources — papers, textbooks, Stack Overflow answers, blog posts, repositories. URL or full citation.
- AI tools — which tool/model, what it was used for (brainstorming, code completion, drafting, rewriting, debugging), and which sections it touched.
- Pre-existing material — own prior work, templates, starter code.
- Where each contribution appears — section, file, line range, figure number.
- Verification — for each AI- or source-derived contribution, can the author explain it without the source? (Rule 3 hook.) Mark
understood / partial / not yet.
Output format
## Honesty Declaration
Course: <code> — <assignment>
Author(s): <name(s)>
Permitted-aids policy: <one-line summary OR "not stated — see note">
### Own work
- <high-level statement of what the author did independently>
### Help received
| Source | Type | Used for | Location | Understood? |
|---|---|---|---|---|
| <name / citation / tool> | person / source / AI / prior work | <purpose> | <where> | yes / partial / no |
### Notes
- <any caveats, e.g. "policy did not address AI use; I treated it as permitted-with-disclosure">
- <anything the author is unsure whether to disclose — flag it rather than hide it>
Signed: <name>, <date>
Rules for the assistant
- Never invent entries the user did not report. If you (Claude) helped with this submission earlier in the conversation, include yourself explicitly — model name, what you did, where it appears.
- If the user under-reports help that is visible in the conversation, point it out before producing the block.
- If
Understood? is no for any row, recommend running honor-defense-prep before submitting.
- If the permitted-aids policy is unknown, recommend
honor-aid-check and produce the block with a note rather than guessing.
