(Proposal, unverified) Grade reproducibility-relevant engineering configuration items (hyperparameter search range, compute budget, seed handling, dataset splits) on a complete/partial/none scale, requiring the grader to first define what "complete" means per item before judging against it. Use this after study-design-tool-gate has dispatched an ML/CS engineering paper here; this is a graded QUALITY judgment, distinct from dual-column-self-check's binary Yes/No/NA self-audit checklists.
SkillScore46Maintained
Automatic
◆QA & Test Engineering pack
Last verified 9/7/2026
What it changes
Passed 10 of 21 test cases with the skill, and 7 of 21 without it.
Without skill33%
With skill46%
Checks it passed that the model alone missed
Identifies NOS as the structural model for the proposal.
Configures the spawned subagent with full MCP tool access.
Includes the string (Proposal, unverified) within the description.
and 1 more in the full report
4 fixed7 passed either way12 still failing1 broke
Pass-rate lift
+13 pp
Tokens
+66%
Agent turns
0%
Response time
+40%
Model gemini-3.6-flashMeasured 9/7/2026v111 of 24 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 4
Prompt
Our team is writing the proposal document for engineering config grading. Which existing operational framework is this method structurally modeled after, despite applying to a different content layer?
What was checked, without and with the skill
Checked
Without skill
With skill
Identifies NOS as the structural model for the proposal.
Missed
Passed
case-13 · gemini-3.6-flash · 9/7/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md21 lines2 files
Engineering Config Grading (Proposal)
Graded (not binary) reproducibility-config quality judgment. Fills the quality-judgment × engineering-metadata gap in the evaluative-stance × content-layer matrix (spec §2, matrix-generation phase). Per coverage-audit M14: an earlier draft folded this into dual-column-self-check via a value-domain toggle alone, which dropped the actual judgment-defining action (establishing what "complete" means) that distinguishes this from a binary checklist.
Execution
Subagent — spawned via spawn-agent skill.
Proposal Status — Read Before Modifying
No primary-source precedent (unlike NOS, which it's structurally modeled after but applies to a different content layer). Keep "(Proposal, unverified)" in the description until real usage validates the method.
Available SOPs
SOP
When to use
spawn-agent
Spawn a customized CC subagent with full MCP tool access.