Across 12 independent cultures, the treatment increased marker expression by 18% (95% CI 10–26%; Fig. 1a). We carefully verified that this was not simply an arbitrary observation. This result should not be interpreted as evidence that the treatment improves cell survival. Survival changed by 1 percentage point (95% CI −3 to 5; Fig. 1b). These results demonstrate that the treatment improves both marker expression and survival.
Across 12 independent cultures, the treatment increased marker expression by 18% (95% CI 10–26%; Fig. 1a). Survival changed by 1 percentage point (95% CI −3 to 5; Fig. 1b), with no clear evidence of improved survival. The treatment therefore increased marker expression without a demonstrated survival benefit.
What the supplied evidence actually says
- The independent unit is a culture: 12 independent cultures, not 12 cells.
- Marker expression: increase of 18%, 95% CI 10–26%; Fig. 1a.
- Survival: change of 1 percentage point, 95% CI −3 to 5; Fig. 1b.
- There is no demonstrated survival improvement. No P value or mechanism is supplied.
- Keep all numbers, units, confidence intervals and figure references.
- No other manuscript sections or editable figure files are available.
- Revise only this Results paragraph. Do not create experiments, citations or extra sections.
An illustrative reference revision, not a recorded live agent execution or a comparative benchmark. “No demonstrated survival benefit” does not mean “no effect.”
Run it in your own agent.
After installation, open this repository in your agent, then use the request below. Inspect your output before reading expected.md; the reference is not an exact answer to copy. Installation guide →
Use paper-workflow. Read examples/first-run/draft.md and examples/first-run/evidence.md. Revise only the Results paragraph using the supplied evidence. Write the revision to examples/first-run/revised.md and briefly explain the material changes. Preserve measured values and figure references.
Preserve the measurements
Keep the independent unit, both effect sizes and confidence intervals, and Fig. 1a/1b. Do not turn cultures into cells or invent a P value.
Correct the unsupported conclusion
The marker-expression finding stays. A demonstrated survival improvement or an untested mechanism must not be written into the paragraph.
Stay within the task
Only one paragraph is in scope. No full submission audit, invented figures or extra experiments are required to complete this tutorial.
Read the original materials
examples/first-run/draft.md
# Synthetic Results excerpt
This fictional example is for learning the workflow. It describes no real study.
Across 12 independent cultures, the treatment increased marker expression by 18% (95% CI 10–26%; Fig. 1a). We carefully verified that this was not simply an arbitrary observation. This result should not be interpreted as evidence that the treatment improves cell survival. Survival changed by 1 percentage point (95% CI −3 to 5; Fig. 1b). These results demonstrate that the treatment improves both marker expression and survival.
examples/first-run/evidence.md
# Synthetic evidence and editing scope
- This is a research article, not a Review.
- The independent unit is a culture: 12 independent cultures, not 12 cells.
- Marker expression: increase of 18%, 95% CI 10–26%; Fig. 1a.
- Survival: change of 1 percentage point, 95% CI −3 to 5; Fig. 1b.
- There is no demonstrated survival improvement. No P value or mechanism is supplied.
- Keep all numbers, units, confidence intervals and figure references.
- No other manuscript sections or editable figure files are available.
- Revise only this Results paragraph. Do not create experiments, citations or extra sections.
examples/first-run/expected.md
# One acceptable revision
This is an illustrative reference output, not a verbatim-output requirement or a recorded model run.
Across 12 independent cultures, the treatment increased marker expression by 18% (95% CI 10–26%; Fig. 1a). Survival changed by 1 percentage point (95% CI −3 to 5; Fig. 1b), with no clear evidence of improved survival. The treatment therefore increased marker expression without a demonstrated survival benefit.
The edit removes reassurance and corrects the unsupported survival claim. It preserves the independent unit, both measured effects, both confidence intervals and both figure references. “No demonstrated benefit” must not become “no effect”: the interval includes changes in both directions.