Calibrates confidence scores based on debate progression. Determines whether to escalate, continue, or terminate based on cumulative evidence.
SkillScore73Maintained
Automatic
◆AI Engineer pack
Last verified 8/22/2026
What it changes
Passed 16 of 22 test cases with the skill, and 6 of 22 without it.
Without skill27%
With skill73%
Checks it passed that the model alone missed
Sets saturation_flag to true
Sets saturation_flag to false
Uses the exact key decision in the output JSON
and 9 more in the full report
10 fixed6 passed either way6 still failing0 broke
Pass-rate lift
+45 pp
Tokens
-18%
Agent turns
0%
Response time
-34%
Model gemini-3.6-flashMeasured 8/22/2026v116 of 22 cases passed with skill
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 have allocated a budget of 5 units for calibrating our distributed lock protocol debate. How many calibration assessments does this budget permit per round?
What was checked, without and with the skill
Checked
Without skill
With skill
States that one unit equals one calibration assessment per round
Missed
Passed
case-12 · gemini-3.6-flash · 8/22/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md40 linessingle file
Confidence Calibration
Calibrates confidence based on debate progression.
Execution
Subagent — spawned via subagent-spawning/spawn-agent.
Why Subagent
Calibration requires meta-analysis of debate trajectory without being anchored to any single round's outcome. Isolated context enables objective trend assessment.
Input
round_verdicts: All judge verdicts so far
confidence_history: Confidence scores from each round
budget_remaining: Rounds/searches remaining in budget
Output
calibrated_confidence: Updated confidence in artifact viability (0.0–1.0)
decision: escalate / continue / terminate
reasoning: Why this decision given the trajectory
saturation_flag: Whether debate is producing diminishing returns
Budget
One unit = one calibration assessment per round.
Available SOPs
Optional, no fixed order; the final leaf is always a sop.
SOP
When to use
spawn-agent
Spawn a customized CC subagent with full MCP tool access. Used by SOPs that declare execution: subagent.