- Home
- Skills
- Technical Documentation
- Document Preflight
Works with the AI Agents you already use
Document Preflight
Technical pre-submission auditor to prevent desk rejections by flagging missing declarations, refs, and figure errors.
Free
Document Preflight
Example session with this skill installed
Run a pre-submission check on my manuscript 'neural_net_final.pdf' for the Journal of Machine Learning. It needs to be anonymized and I'm worried about the figure captions and refs.
- Read your context and instructions
- Compiled the document preflight
I have completed the preflight check for your manuscript against the Journal of Machine Learning standards.
Preflight — neural_net_final.pdf for Journal of Machine Learning
Verdict: NOT READY · 2 BLOCK · 3 FIX · 6 PASS
| Item | Evidence | Fix |
|---|---|---|
| BLOCK | ||
| Reference [14] | In-text citation exists but missing from bibliography | Add the full citation for Smith et al. (2022) |
| Figure 2 Caption | Placeholder text "Figure 2 goes here" | Replace with a descriptive caption explaining the architecture |
| FIX | ||
| Abstract length | 285 words detected | Shorten to under the 250-word limit |
| File naming | neural_net_final.pdf | Rename to 'manuscript_anonymized.pdf' per JML guidelines |
| Table 1 | Missing units for 'Latency' column | Add (ms) to the column header |
PASS
Anonymization (no author names found), Keywords, Ethics Statement, Data Availability Statement, Figure 1 Labels, Title Page.
Next steps
- Resolve the missing reference in the bibliography and update the Figure 2 caption.
- Trim the abstract by 35 words to meet the technical limit.
- Once these blocks are cleared, you can run
academic-manuscript-formatterto prepare the final PDF.
Connects securely to your tools. The creator never sees your data.
About this skill
Submission-ready documents require more than just good writing. Formatting errors, missing declarations, and broken references often lead to immediate desk rejections or technical returns. This skill scans manuscripts, theses, and reports to identify these blockers before you hit submit.
What it does
- Compliance Audit checks for missing declarations, ethics statements, and required front matter.
- Reference Validation identifies unresolved citations and missing bibliography entries.
- Asset Review flags unlabeled figures, missing table units, and vague captions.
- Anonymization Check ensures documents intended for double-blind review are properly scrubbed.
- Metadata Verification audits file naming conventions and abstract word counts against target guidelines.
How it works
- Load Checklist uses the
submission-checklist.mdfor the specific document type: journal manuscript, thesis, or report. - Evidence Gathering scans the provided file or section list for specific elements required by the target venue.
- Status Grading assigns a
PASS,FIX, orBLOCKstatus to each requirement based on direct evidence. - Prioritized Reporting generates a structured list of issues, ranking critical blockers that require author decisions at the top.
Frameworks & tools
Works with Word documents, LaTeX files, and Markdown manuscripts. Uses internal checklists for academic and professional submission standards.
Why this beats prompting it yourself
Generic prompts often hallucinate journal requirements or miss technical details like figure/table consistency. This skill uses a structured grading system and specific evidence-based checks to ensure no assumption is marked as a pass.
Use cases
- Auditing a journal manuscript before final submission to avoid technical desk rejection.
- Verifying thesis front matter and chapter order against university graduation requirements.
- Checking a grant proposal for mandatory data availability and ethics statements.
- Reviewing report versioning and approval logs for professional distribution.
Known limitations
Does not format, rewrite, or edit prose. Cannot submit documents to portals on behalf of the user. Requires specific target guidelines to be provided for custom venue rules.
Does not reformat or rewrite content. Requires user-provided guidelines for custom venue rules. Cannot verify the truth of data statements, only their presence.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
This skill has been used 2 times.
Trust & safety
Security scan
Security scan passed 27 days ago
- Free with an Agensi account
Permissions
File Scopes