SuperAPI AI Directory

byted-sol-stability-sli-modeling

Deployment & operations

View repository

Updated 10-15 17:30

About this skill

SLI Modeling Skill

Input

  • Capability/scenario description text (required)
  • owner (required)
  • Reference document path (optional)

Output

Fixed output to output/<slug>/:

  • sli-spec.json
  • sli-report.md

SLI Spec fields (hard constraints)

  • capability
  • user_journey
  • sli_name
  • sli_type:availability|latency|correctness|freshness|completeness|consistency
  • measurement
  • denominator
  • dimension
  • target_slo
  • error_budget
  • severity:P0|P1|P2
  • owner

Execution rules

  1. Must not output an SLI Spec with missing fields.
  2. sli_type and severity must match the enumeration.
  3. Field validation failures must return a diagnosable error, and must not silently fill in meaningless default values.
  4. It is allowed to explicitly specify fields and override inferences via the key: value form in the input text.
  5. Prioritize modeling around critical user journeys (such as login, core requests, checkout), and avoid using internal resource metrics such as CPU/memory directly as SLIs.
  6. Prefer the request-based approach (good requests / total requests); adopt the period-based approach only when truly necessary, and explicitly state the window in target_slo .
  7. Use a rolling 30d target window by default,error_budget by default follows the 1 - target_slo approach.
  8. Do not use 100% as the default target; it is recommended to use a 99.x target and observe budget consumption via burn-rate.

CLI

byted-sol-stablity-sli-modeling \
  --input examples/input.capability.md \
  --owner team-observability \
  --out-dir output