- Home
- Skills
- Research & Analysis
- Lab Meeting Report
Works with the AI Agents you already use
Lab Meeting Report
Turn research notes, papers, and experiment results into evidence-grounded lab meeting reports.
Free
Lab Meeting Report
Example session with this skill installed
Use lab-meeting-report to read weekly-notes.md, results.csv, and paper-notes.md in examples/weekly-summary.
Create a concise English lab meeting report for a 10-minute discussion. Include this week's progress, a results table, failed experiments, blockers, and next actions. Preserve source references and distinguish observed results from untested ideas.
- Read your context and instructions
- Compiled the lab meeting report
- Generated the data export
Weekly Lab Meeting Report: Microscopy Segmentation
Synthetic demonstration. All project details and results are fictional.
Reporting period: July 6–12, 2026 | Pre-meeting report
Weekly Overview
Progress: Completed three seeds with the current U-Net configuration. Added stain normalization and fixed the data loader's random state. All three runs completed training.
Sources: weekly-notes.md, results.csv
Key evidence: Dice scores were 0.812, 0.806, and 0.809 for seed-11, seed-23, and seed-37 respectively.
Source: results.csv
Discussion needed: Physical batch size 4 caused an out-of-memory failure on a 16 GB GPU despite AMP and cache clearing. Decide whether to borrow a GPU with at least 24 GB or use gradient accumulation.
Sources: weekly-notes.md, results.csv
Previous Action Review
| Action | Status | Artifact |
|---|---|---|
| Complete three seeds using the current configuration | Complete | results.csv |
Sources: weekly-notes.md, results.csv
Experimental Results
| Configuration | Seed | Batch size | Dice | IoU | Peak GPU memory | Status |
|---|---|---|---|---|---|---|
| Current | seed-11 | 2 | 0.812 | 0.691 | 14.8 GB | Complete |
| Current | seed-23 | 2 | 0.806 | 0.683 | 14.9 GB | Complete |
| Current | seed-37 | 2 | 0.809 | 0.687 | 14.8 GB | Complete |
Source: results.csv
These observations confirm that all three runs completed. The supplied material does not establish statistically significant improvement.
Failed Attempts and Evidence Limits
Boundary dropout: Dice was 0.774 for seed-23, compared with 0.806 for the current configuration using the same seed. This week's evidence does not support retaining this branch.
Physical batch size 4: Training failed at the first optimizer step on the 16 GB GPU. AMP and cache clearing did not resolve the failure.
Sources: weekly-notes.md, results.csv
Literature idea: The boundary-aware auxiliary loss in the paper notes was studied on fluorescence microscopy. Its applicability to the current bright-field data is unknown, and it has not been implemented in this project.
Source: paper-notes.md
Blocker and Requested Decision
The current GPU cannot complete training with physical batch size 4.
Choose between
- Borrowing a GPU with at least 24 GB to test physical batch size 4.
- Keeping physical batch size 2 and using gradient accumulation for effective batch size 4.
Source: weekly-notes.md
Next Actions
| Action | Deliverable | Completion criterion |
|---|---|---|
| Run the batch configuration experiment using the agreed route | Training logs and a comparison table for three seeds | All runs finish without OOM; record Dice, IoU, and peak GPU memory |
| Assess the boundary-aware auxiliary loss | A documented keep-or-defer decision | Use the current configuration as comparator; do not treat paper results as results on this dataset |
If no 24 GB GPU is available, proceed with gradient accumulation. No performance improvement threshold has been predefined.
Sources: weekly-notes.md, paper-notes.md
Sources
weekly-notes.mdresults.csvpaper-notes.md
results.csv
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Prepare your next lab meeting from the research material you already have.
Turn weekly notes, papers, experiment results, figures, and code changes into a concise Markdown report covering progress, evidence, blockers, and next steps.
- Create weekly research summaries, progress reports, and journal club notes.
- Preserve failed experiments, negative results, and uncertainty.
- Keep facts, interpretations, and hypotheses distinct.
- Avoid invented metrics, citations, and explanations.
- Generate reports in Chinese or English.
- Use optional guidance for presentation preparation and Feishu/Lark or Notion publishing.
Getting started
“Use lab-meeting-report to read my weekly notes, results, and paper notes, then prepare a report for a 10-minute lab meeting.”
Open source under the MIT license.
No DOCX support. Requires explicit scope for cloud integration. Will not search the web for literature unless asked. Manual cleanup needed for sparse inputs.
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
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires