Use when experiments complete to judge what claims the results support, what they don't, and what evidence is still missing. Codex MCP evaluates results against intended claims and routes to next action (pivot, supplement, or confirm). Use after experiments finish — before writing the paper or running ablations.
SkillScore77Maintained
Automatic
◆Research pack
Last verified 8/11/2026
What it changes
Passed 15 of 19 test cases with the skill, and 8 of 19 without it.
Without skill41%
With skill77%
Checks it passed that the model alone missed
Proceeds directly to paper writing.
Specifies gpt-5.4 as the reviewer model.
Extracts the suggested_claim_revision field.
and 7 more in the full report
9 fixed8 passed either way4 still failing1 broke
Pass-rate lift
+36 pp
Tokens
+13%
Agent turns
0%
Response time
-56%
Model gemini-3.6-flashMeasured 8/11/2026v217 of 22 cases passed with skill · 19 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 9
Prompt
If the evaluation suggests reframing or weakening an overextended claim, which parsed field holds that recommendation?
What was checked, without and with the skill
Checked
Without skill
With skill
Extracts the suggested_claim_revision field.
Missed
Passed
case-15 · gemini-3.6-flash · 8/11/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md123 linessingle file
Result-to-Claim Gate
Experiments produce numbers; this gate decides what those numbers mean. Collect results from available sources, get an objective judgment, then route based on the verdict.
Context: $ARGUMENTS
Constants
REVIEWER_MODEL = gpt-5.4 - Used via a secondary Codex agent for objective claim assessment.
When to Use
After a set of experiments completes (main results, not just sanity checks)
Before committing to claims in a paper or review response
When results are ambiguous and you need an objective second opinion
Workflow
Step 1: Collect Results
Gather experiment data from whatever sources are available in the project:
W&B (preferred): wandb.Api().run("<entity>/<project>/<run_id>").history() - metrics, training curves, comparisons
EXPERIMENT_LOG.md - full results table with baselines and verdicts
EXPERIMENT_TRACKER.md - check which experiments are done vs still running
Log files - ssh server "tail -100 /path/to/training.log" if no other source
docs/research_contract.md or project notes - intended claims and experiment design
Assemble the key information:
What experiments were run (method, dataset, config)
Main metrics and baseline comparisons (deltas)
The intended claim these experiments were designed to test
Any known confounds or caveats
Step 2: Secondary Codex Judgment
Send the collected results to a secondary Codex agent for objective evaluation: