AB self-improving-ai
Captures learnings about GenAI/LLM configuration, model selection, inference optimization, fine-tuning, RAG pipelines, prompt engineering, multimodal processing, and cost management. Use when: (1) Model response quality degrades after a provider update or version change, (2) Inference latency exceeds acceptable thresholds, (3) Fine-tuned model regresses on evaluation benchmarks, (4) RAG retrieval returns irrelevant or stale chunks, (5) Token costs exceed budget projections, (6) Hallucination rate increases on factual queries, (7) Context window overflows cause critical information truncation, (8) Multimodal pipeline fails on specific input types (image, audio, video, PDF), (9) A better model or configuration is discovered for a task, (10) Guardrails block valid output or miss harmful content.
As a process B 66/100 · Nearly there — weak spots: 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 ≈ 7157 tokens (recommended < 5000); move details to references/
Process rating: all ten parameters 66/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 30Running it twice. 18 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
- 60Result and completion. Output format stated, no completion criterion
- 70Execution cost. Instruction body is 7157 tokens
- 100Steps. 112 steps
- 100Failures and branches. 2 branches, has a failure section
- 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 21 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)
- +3Description length 803: 120–800 characters recommended
- +1No license
- +2Single-language instructions
- +4Structure: 53 headings
- +3Step-by-step instructions: 112 items
- +3Output format is stated explicitly
- +4Has examples (18 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.