disclosures

name	disclosures
description	Generate an AI-use disclosure, CRediT authorship statement, and/or funding statement for the manuscript — the boilerplate sections most journals now require at submission. Triggers on: 'AI disclosure', 'author contributions', 'CRediT statement', 'funding statement', 'who contributed what', 'acknowledgments section'.
disable-model-invocation	true
argument-hint	<ai|credit|funding|all>

/disclosures — AI-Use, Authorship, and Funding Statements

Templates three boilerplate statements journals increasingly require at submission: an AI-use disclosure, a CRediT (Contributor Roles Taxonomy) authorship statement, and a funding statement. All three are produced by scripts/compliance_gen.py, which never invents a fact — every mode requires its inputs explicitly (which AI tool, what it was used for, which authors did what, which grants funded the work). If you don't have an input, ask the user rather than guessing a plausible-sounding default; a wrong funding source or an uncredited author is a real, reportable integrity problem, not a cosmetic one.

Procedure

1. Determine Which Statement(s) to Generate

Argument ai, credit, funding, or all (default all if omitted — ask the user to confirm scope rather than silently generating all three if the request's intent was ambiguous).

2. AI-Use Disclosure

1. Ask (if not already known from context): which AI tool(s) were used, and for what specific tasks (e.g., "literature synthesis," "citation formatting," "grammar checking") — not a vague "AI was used." Also ask who verified the citations/claims (usually the corresponding author).
2. Ask which venue-family framing fits: icmje (medical/health journals), ieee (engineering/CS), or generic (no specific framing). This changes only the opening sentence, not the substantive content.
3. Run:

   python scripts/compliance_gen.py ai-disclosure --tool "<tool>" --uses "<uses>" \
     --venue-family <family> [--verified-by "<name>"]
   

4. Always relay the script's warning field to the user — this statement is a starting point, not a guarantee that it satisfies the target venue's exact required wording. Point back to /submit-check Step 7's existing reminder to verify against the actual venue policy.

3. CRediT Authorship Statement

1. Ask for the full author list and, for each author, which of the 14 official CRediT roles they performed: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.
2. Do not assume a role from context (e.g., "you probably did Writing – original draft since you're the corresponding author") — ask, or confirm an inferred guess explicitly before including it.
3. Run:

   python scripts/compliance_gen.py credit --authors \
     '[{"name":"Jane Doe","roles":["Conceptualization","Software"]}, ...]'
   

   An unrecognized role name is rejected by the script (exit 1) rather than silently coerced — if that happens, ask the user which of the 14 official roles they meant.
4. The unassigned_roles field lists CRediT roles no author was assigned — informational only; not every project uses every role, but flag it in case a contribution (e.g., Funding acquisition) was simply forgotten.

4. Funding Statement

1. Ask whether the work received external funding.
   • If yes: for each funding source, ask the funder name, grant/award number (if any), and which author(s) it was awarded to.
   • If no: confirm explicitly — do not default to "no funding" just because the user didn't mention any; ask the direct question.
2. Run:

   python scripts/compliance_gen.py funding --grants \
     '[{"funder":"...", "grant_number":"...", "recipient":"..."}]'
   # or, for explicitly no funding:
   python scripts/compliance_gen.py funding --none
   

5. Write the Output

Append (or create) drafts/compliance-statements.md with one section per statement generated:

## AI Disclosure
<statement text>

## Author Contributions (CRediT)
<statement text>

## Funding
<statement text>

This is a supplementary file, not one of the hook-protected data files — write it directly with the Edit/Write tool, same as any other drafts/*.md file. Tell the user these sections belong in the manuscript's Acknowledgments/ Author Contributions/Funding front- or back-matter per their target venue's own placement convention (format-engineer places them at build time — this skill only generates the text, it does not itself build the manuscript).

GATES SUMMARY

Gate	Trigger	Action
AI tool or specific uses not known	Yes	ASK, don't guess a plausible-sounding default
Author role not confirmed by the user	Yes	ASK or explicitly confirm an inferred guess before including it
Funding status not stated by the user	Yes	ASK explicitly — never default to "no funding"
Unrecognized CRediT role name	Yes	Script rejects (exit 1); re-ask for one of the 14 official roles

See Also

• scripts/compliance_gen.py --help — the three subcommands and their exact flags.
• .claude/skills/submit-check/SKILL.md Step 7 — the existing AI-disclosure reminder this skill's output feeds into.