Use when adding or auditing Agentlas skill lifecycle metadata, skill-registry.json, trial evidence, Curator promotion decisions, or first-class skill promotion gates.
SkillScore86Maintained
Automatic
◆AI Engineer pack
Last verified 8/15/2026
What it changes
Passed 18 of 21 test cases with the skill, and 13 of 21 without it.
Without skill64%
With skill86%
Checks it passed that the model alone missed
Sets initial export tier to candidate
Requires situation tags for every skill
Includes evidence gaps in report output
and 7 more in the full report
5 fixed14 passed either way3 still failing0 broke
Pass-rate lift
+23 pp
Tokens
-21%
Agent turns
0%
Response time
-42%
Model gemini-3.6-flashMeasured 8/15/2026v119 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 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 5
Prompt
What required output elements must be included in the final governance report when performing a skill promotion audit on an Agentlas repository?
What was checked, without and with the skill
Checked
Without skill
With skill
Includes evidence gaps in report output
Missed
Passed
Includes residual risks in report output
Missed
Passed
case-17 · gemini-3.6-flash · 8/15/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md29 linessingle file
Skill Lifecycle Promotion
Use this skill when a generated or packaged Agentlas repo needs governed skill
promotion metadata.
Procedure
Add .agentlas/skill-registry.json as an export-only candidate registry.
Add empty .agentlas/skill-trials.jsonl and
.agentlas/curator-decisions.jsonl.
Keep every skill at tier: candidate on export.
Keep runtimeFirstClassRecallEnabled: false unless a local Curator later
approves promotion.
Add success predicates and situation tags for every skill.
Separate ## Memory Events from ## Skill Trial Events.
Treat LLM rubric review as weak evidence only.
Block promotion when authority separation, sealed holdouts, replayability, or
rollback evidence is missing.
Include false-accept, blind-spot, and drift terms in any durable-error budget.