AC self-improving-science
Captures learnings, experiment issues, and methodology corrections for continuous improvement in scientific research and ML workflows. Use when: (1) Data leakage detected in train/test split, (2) Model fails to reproduce across seeds or environments, (3) Statistical test misapplied or p-value misinterpreted, (4) Hypothesis test fails or needs revision, (5) Feature distribution shift detected, (6) User corrects methodology or analysis approach, (7) Experiment design flaw discovered. Also review learnings before designing new experiments.
As a process C 51/100 · Has gaps — weak spots: result and completion, inputs and preconditions, running it twice
How to improve
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 15. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
body-longSKILL.md body ≈ 5673 tokens (recommended < 5000); move details to references/
Process rating: all ten parameters 51/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 30Running it twice. 19 mutating operations with no state check
- 50When it triggers. No condition that starts the skill
- 60Tools and files. Uses tools (bash) that frontmatter does not declare
- 65Failures and branches. 3 branches
- 70Execution cost. Instruction body is 5673 tokens
- 85Steps. 103 steps, 1 vague phrases
- 100Consistency. Name and required fields are in place
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 19 top-level sections: this looks like several domains in one skill
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +3Description length 542: enough signal without eating the budget
- +4Structure: 44 headings
- +3Step-by-step instructions: 103 items
- +4Has examples (21 code blocks)
- +4Reference files are cited in the instructions (2 of 3)
- +3All 3 scripts are documented
Quality base 70; lint remarks subtract, signals add up to 100. Result: 81.