agentsclimarketplace

Sci evidence discussion

Skill keithhegit/sci-using-kit/skills/sci-evidence-discussion

A skill family for evidence-bounded SCI manuscript workflows: literature vetting, claim-evidence alignment, paper revision, reviewer preflight, LaTeX editing, and submission packaging.

Install
npx -y skills add keithhegit/sci-using-kit --skill sci-evidence-discussion

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when converting experiment tables, logs, figures, measurements, HIL outputs, simulations, or appendix data into results, discussion, limitations, or contribution text

SKILL.md

1.4 KB, as published. Nobody here has run it

SCI Evidence To Discussion

Overview

Turn measured evidence into manuscript discussion without overclaiming.

Core principle: separate observation, interpretation, and claim strength.

Evidence Table

Use or update the project's experiment evidence table when present. Otherwise record the fields below in the response.

Record:

  • metric name and unit
  • source file/log/table
  • sample count or run scope
  • baseline/reference definition
  • measurement environment
  • evidence level
  • caveats and exclusions

Discussion Pattern

  1. Observation: what the data show.
  2. Mechanism: why this may happen, only if supported.
  3. Scope: where the result applies.
  4. Boundary: what the result does not prove.
  5. Next evidence: what would strengthen the claim.

Output

  • results paragraph
  • discussion paragraph
  • limitation sentence if needed
  • claim-evidence mapping row

Common Mistakes

  • Confusing detector-only latency with end-to-end agent latency.
  • Treating manual power-meter screen readings as calibrated power traces.
  • Moving Paper 2/downlink results into Paper 1 without scope separation.
  • Reporting steady-state results as all-phase behavior.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.