Skill Cold Start Warm Path Optimizer

Skill Cold-Start Warm-Path Optimizer

Use this skill to convert cold-start latency into deterministic warm-path planning.

Workflow

1. Collect execution logs with timestamp, skill, duration, and optional cold/cache fields.
2. Run warm_path_optimizer.py analyze for cold vs warm metrics.
3. Generate a prewarm and auto-invoke policy with warm_path_optimizer.py plan.
4. Decide whether prewarm is required for this chain and record the decision.
5. Apply the plan before high-throughput sessions.

Analyze

python3 "$CODEX_HOME/skills/skill-cold-start-warm-path-optimizer/scripts/warm_path_optimizer.py" analyze \
  --input /path/to/skill-latency.csv \
  --window-days 30 \
  --cold-penalty-min-ms 800 \
  --min-invocations 3 \
  --rare-skill-max-invocations 2 \
  --never-auto-penalty-ms 3000 \
  --json-out /tmp/cold-warm-analysis.json \
  --format table

Build Plan

python3 "$CODEX_HOME/skills/skill-cold-start-warm-path-optimizer/scripts/warm_path_optimizer.py" plan \
  --analysis-json /tmp/cold-warm-analysis.json \
  --max-prewarm 10 \
  --json-out /tmp/cold-warm-plan.json \
  --format table

Mandatory Chain Gate

Before finalizing skill chains, provide:

• cold_warm_analysis_json=/tmp/cold-warm-analysis.json
• cold_warm_plan_json=/tmp/cold-warm-plan.json
• prewarm_required=<true|false>

If this evidence is missing, chain selection is incomplete and must fail closed.

Prewarm Decision Rubric

Set prewarm_required using deterministic thresholds:

1. true when:
   • cold penalty exceeds configured threshold, and
   • invocation frequency is above rare-skill threshold, and
   • lane is latency-sensitive for current task.
2. false when:
   • skill is rare in window, or
   • cold penalty is below threshold, or
   • lane is non-latency-critical.

Also record never_auto_invoke skills from the plan output to prevent hidden churn.

Scope Boundary

Use this skill only for the skill-cold-start-warm-path-optimizer lane and workflow defined in this file and its references.

Do not use this skill for unrelated lanes; route those through $skill-hub and the most specific matching skill.

References

• references/optimizer-workflow.md
• references/latency-contract.md

Loopback

If this lane is unresolved, blocked, or ambiguous:

1. Capture current evidence and failure context.
2. Route back through $skill-hub for chain recalculation.
3. Resume only after the updated chain returns a deterministic next step.