Create or improve portable agent skills with clear trigger metadata, focused instructions, bundled resources, safety boundaries, and real validation.
SkillScore91Maintained
Automatic
◆AI Engineer pack
Last verified 8/3/2026
What it changes
Passed 19 of 21 test cases with the skill, and 11 of 21 without it.
Without skill50%
With skill91%
Checks it passed that the model alone missed
does not create a SKILL.md file
includes a Core Judgment heading
creates a skill.yml file for registry metadata
and 13 more in the full report
11 fixed9 passed either way0 still failing2 broke
Pass-rate lift
+41 pp
Tokens
+67%
Agent turns
0%
Response time
-9%
Model gemini-3.6-flashMeasured 8/3/2026v120 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 10
Prompt
We finished building an agent skill for API contract testing. What specific report elements must be included in the concluding summary response?
What was checked, without and with the skill
Checked
Without skill
With skill
includes validation checks performed in the final report
Passed
Passed
includes open risks or remaining questions in the final report
Missed
Passed
case-14 · gemini-3.6-flash · 8/3/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md307 linessingle file
Skill Authoring
Use this skill to create, improve, review, or package an agent skill.
A good skill gives a capable agent the missing procedural knowledge it needs to do a repeatable job well. It should not micromanage obvious reasoning, restate generic best practices, or bury the agent in ceremony.
Core Judgment
Write for an intelligent agent with limited context.
The skill should answer:
what reusable capability this adds
when the agent should load it
what source material or tools it needs
what workflow or judgment it should follow
what outputs or evidence prove the work is complete
what actions are unsafe without explicit user approval
If a section would not change agent behavior, remove it.