Use This Page Only When a Regular Prompt Is Not Enough.

I ran an OER or zero-cost course-material search and the results were not useful.

Before searching again, ask me for:
1. the course or topic,
2. what I was trying to find,
3. the search terms or prompt I used,
4. examples of results that were useful or not useful, and
5. any important constraints.

Then diagnose the likely problem with the search. Consider terminology, scope, discipline, resource type, overly broad criteria, overly narrow criteria, weak source selection, and whether I may be asking for something that is rarely available as one resource.

Propose 3 to 5 specific changes to the search. If the original search was already reasonable and the real problem is limited availability, say so instead of inventing a more complicated strategy.

If you have current web access, carry out the revised search only after showing me the plan. If you cannot access current source pages, give me the search terms and source types to try and do not claim that resources, links, licenses, availability, or accessibility information are verified.

Do not invent resources, links, licenses, availability, or accessibility information. Separate verified facts from inference and label unresolved items “Needs verification.” Keep the final candidate list manageable.

I have an AI prompt for OER, ZTC, course-material, or faculty-support work that I want to improve.

Ask me to paste the prompt first. Then ask no more than 3 short questions needed to understand:
- what I actually want the prompt to accomplish,
- what has been going wrong, and
- what information or judgment must remain human.

After I answer, identify the smallest changes that would make the prompt work better. If one or two edits are enough, do not rewrite the entire prompt.

Preserve my actual goal rather than making the task broader or more elaborate. Build in these safeguards when relevant:
- do not invent titles, links, licenses, holdings, or access conditions;
- separate evidence from inference;
- label unknowns “Needs verification”;
- do not make final curricular, licensing, accessibility, or adoption decisions for me;
- keep the output as short as the task reasonably allows.

Show me:
1. the revised prompt,
2. up to 3 brief changes you made and why, and
3. anything the prompt still cannot safely or reliably do.

I have AI-generated research about OER or zero-cost course materials that I need to audit before I use it.

Ask me to paste the research output, list, table, or memo. Then review the important factual claims.

For each important claim, use one status:
- Verified: confirmed from an original or authoritative source,
- Needs verification: plausible or incomplete but not yet confirmed, or
- Not verified: unsupported, contradicted, or not checkable with the information available.

Use the Evidence column to explain why a claim received that status, including whether it was contradicted or simply could not be checked.

Pay special attention to:
- whether a resource actually exists,
- the correct title and source,
- direct links,
- open-license or public-domain status,
- whether “free to access” has been confused with OER,
- library or subscription claims,
- accessibility claims,
- currency and version information, and
- claims about course fit that are not supported by course evidence.

If you have current web access, prefer original source pages. If you cannot access the current authoritative source for a claim, do not call it verified. Do not replace one unsupported claim with another.

Return a short audit table: Claim or Candidate | Status | Evidence | What I Should Check Next. End with no more than 3 items that most need human attention.

Help me decide whether an OER, ZTC, course-material, or faculty-support process is worth turning into a reusable workflow.

Ask me first to describe the task, how often I repeat it, and one example of how I currently do it.

If the task is occasional, highly context-specific, or faster to do directly, tell me not to formalize it. If a reusable workflow would genuinely save time, ask only the questions needed to identify:
- who will use it,
- the minimum information they need to provide,
- which repetitive steps AI can assist with,
- which steps require human or local institutional judgment,
- what evidence must be verified, and
- what a useful final output should look like.

Then create:
1. a short step-by-step workflow,
2. a reusable AI prompt that asks for missing information instead of relying on bracket placeholders,
3. a short Human Check list, and
4. a “Needs verification” list for claims the AI should never guess.

Create a separate Quick version only if the full workflow is genuinely more complex than most users will need.

Keep the workflow practical for busy community college faculty or staff. Do not add steps merely to make the process look comprehensive.