Availability and Requirements
The M&E Review ChatGPT edition contains twelve review methods in package me-review-openai, version 0.1.4. Installation of twelve skills has been observed in one personal ChatGPT account. Public installation is not available from this page yet. Actual pasted-text reviews have been run for all twelve methods, and specific upload checks are recorded below. Review and revision journeys were inspected using fictional materials in one personal account, with core tests on version 0.1.2 and targeted repairs checked on versions 0.1.3 and 0.1.4. Some recovery checks shared a bounded chat context. This does not establish every account, file or future response.
Your OpenAI account access, workspace permissions and usage limits apply. The review package adds no M&E Studio signup, purchase, API key or account-linking requirement. The broader workspace drafting and planning skills remain separate.
Installation and First Use
- Check whether your ChatGPT account or workspace permits the plugin. One personal account has been checked; broader account and workspace eligibility remains unverified.
- When a verified installation destination is available, confirm the M&E Review developer and release and follow the controls shown in that account. A private test link does not establish public availability. In the tested personal account, the plugin page offered Try in chat. In a chat, Add files and more showed M&E Review.
- Start a chat with the installed plugin selected. Choose the review that fits the document and explain the purpose. The starters on task pages are for use with that installed method.
- Begin with the fictional logframe example below. Paste readable text with a version name. Treat every proposed result or numeric value as something to verify against your source.
- Inspect the review scope, located findings, strengths, limitations and requested evidence. Resolve unreadable or missing sections before relying on a full-document assessment.
- Make the changes you choose in your own document, supply a named revised version and ask for a re-review against the previous findings.
For current platform guidance, see OpenAI plugin installation and testing documentation. Broader account eligibility and a public installation destination remain unverified.
Readable Inputs and Output Limits
Upload checks apply to the specific review, file and tested personal account listed below. Successful extraction does not establish review quality or support for every document in the same format. Confirm extracted headings, row IDs and causal links before relying on a review. A scan or inaccessible file requires a readable alternative and an explicit scope limit.
| Review | Input Format | Observed Behavior | Limits |
|---|---|---|---|
| Logframe | DOCX | Thirteen paragraphs read and supplied framework content located. | One supplied logframe reviewed in version 0.1.4 with consistent ratings and totals. Other DOCX files remain unverified. |
| Logframe | Text PDF | One page read and supplied framework content located. | One supplied logframe reviewed in version 0.1.2. Scanned PDFs and other PDF documents remain unverified. |
| Indicator quality | XLSX | Indicators worksheet and actual indicator cells read. | One supplied indicator table reviewed in version 0.1.2. Other worksheets and workbook features remain unverified. |
| Theory of change | PNG diagram | Readable diagram content extracted and inspected. | One supplied diagram reviewed in version 0.1.3. Provide a text alternative; other images remain unverified. An actual empty PNG prompted requests for readable evidence across all twelve methods in one shared test chat. |
Keep row IDs, headings, indicator labels and table columns visible. Provide only the sections needed and state what is missing. Copy or download a result only through controls that actually work in your ChatGPT surface. The tested account's Copy response control copied a review successfully. Separate Word, PDF and spreadsheet exports were not verified; the instructions alone do not create a saved file.
Review a Revised Document
After the first review, supply version 2 with the earlier findings. Ask: Review this revised version against the prior findings. Distinguish resolved, unresolved and newly introduced issues, and identify the evidence still needed.
In a new chat, supply both documents or the revised document and previous review. Cross-chat memory is not promised. If the prior review is unavailable, ask for a fresh assessment without claiming an issue was resolved.
Your Document and MeStudio
Provide only material you are authorized to share with ChatGPT. Your document is handled in your ChatGPT session under the settings and terms that apply to that account. The plugin does not automatically send your document to M&E Studio.
You control each transfer: copy or export work from an available tool, supply it in ChatGPT yourself, then decide which suggestions to apply manually. There is no automatic sync or shared saved-work connection. MeStudio transfer formats and receiving behavior are not accepted for this edition yet.
The existing Claude setup route remains separate. Its commands and checking evidence do not establish ChatGPT behavior.
Troubleshooting
- Plugin unavailable
- Check the exact plugin identity and your account or workspace permissions. Public installation is not available from this page yet. Do not follow the Claude installation commands in ChatGPT.
- Workspace installation disabled
- Ask your workspace administrator about permitted installation. The plugin cannot bypass account policy.
- Wrong review selected
- Name the document type and intended review, and request that review only. If the purpose is unclear, clarify it before proceeding.
- Attachment inaccessible
- Paste a short labeled excerpt or supply a readable alternative. Request a bounded review of what is actually readable.
- Scan, diagram or table unreadable
- Provide a transcription, clearer version or labeled rows with column headings. Confirm extracted labels and causal links before relying on the review.
- Document too large or truncated
- Split it into labeled sections with a version and outline. Keep the review scope explicit; do not treat a partial review as a full-document assessment.
- Previous chat unavailable
- Supply the revised document and earlier review together. If the review is unavailable, request a fresh review and state that change tracking cannot be established.
- Download fails
- Use an available copy control. Do not rely on a claimed file without an accessible download. Word, PDF and spreadsheet export are not promised.
- Source version mismatch
- Check the installed plugin release against this guide. Repeat affected reviews after updating through the verified installation route.
- Ratings or totals conflict
- Pause use of the rating. Ask for the summary and final labels to be reconciled against this review method's criteria and any applicable score calculation. Keep located findings and unresolved evidence visible. If the conflict remains, treat the rating as unresolved and request a fresh review with the plugin release named.
- Unverifiable citation or number
- Treat it as unverified. Supply the source, request a correction and keep unresolved evidence gaps visible.
Try a Fictional Review
This is an authored illustration, not a recorded ChatGPT result or practitioner trial. Choose another of the twelve review tasks if it better fits your document.
Use This Review in ChatGPT
ChatGPT edition 0.1.4: a personal-account installation has been observed. Actual review and revision checks have been inspected within the account, version and file limits described in the setup guide. The starter below prepares a request for the installed method. Verify findings against your supplied evidence.
- What It Checks
- Check vertical and horizontal logic, indicators, assumptions and verification.
- What to Provide
- Supply the logframe with row labels, program context and indicator definitions.
- What Comes Back
- A scoped review using this method's own structure, with strengths, located findings, priorities, limitations and proposed improvements.
Select and Copy Manually
Fictional Example and Starter
Review this logframe for results logic and measurement. Locate findings by row and identify the information needed. Use only readable supplied evidence, preserve the method's rating rules and label limitations.
Fictional input: Fictional River Skills v1: Outcome O1: 12 training sessions delivered. Indicator I1: attendance count. Goal G1: sustained use of safer practices. No adoption measure or assumptions supplied.
Illustrative finding: O1 and I1 describe delivery; they do not demonstrate the intended change in practice. Clarify the result and its measurement without inventing a target.
This example was written by hand. It is not an actual ChatGPT output or a test result.
Decide which changes to make. Supply a named revised version and the prior review for a recheck.
Related Rubrics
Open a rubric as a separate reference. Check its scope before copying it into a chat; its rating scale may differ from this review method. You choose what to transfer.