Retry strategy data simulation
Skill kjuhwa/skills-hub/skills/workflow/retry-strategy-data-simulation
Self-correcting knowledge corpus for Claude Code — 9 stable shape clusters, bias-correction pipeline baked into contribution flow. 47 papers, 45 techniques, 1.1k skills.
npx -y skills add kjuhwa/skills-hub --skill retry-strategy-data-simulationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Generate synthetic failure streams and retry attempts to stress-test backoff, jitter, and budget policies
SKILL.md
2.0 KB, 352 tokens by cl100k_base, as published. Nobody here has run it
retry-strategy-data-simulation
Drive the simulation from a configurable failure model rather than random booleans: define per-endpoint failure probability, failure mode distribution (timeout vs 429 vs 503 vs connection-reset), and optional burst windows where probability spikes to simulate upstream incidents. Each tick, spawn new requests at a target RPS, and for every in-flight request evaluate the strategy under test: compute next delay from the policy (fixed/linear/exponential with full/equal/decorrelated jitter), decrement the retry budget, and check circuit-breaker state. Emit an event log of {requestId, attempt, scheduledAt, firedAt, outcome, delayMs, budgetRemaining} that feeds both the timeline visualization and the metrics panel.
Make the failure generator deterministic given a seed so users can replay scenarios and compare strategies side-by-side: same seed + same failure model, swap only the retry policy, and diff the resulting success rate and tail latency. Precompute the next-delay schedule when a request is first queued rather than recomputing on each retry, which keeps the simulation cheap at high RPS and lets the UI draw the full projected track before attempts actually fire. Cap simulation memory by evicting completed requests older than the visible time window into an aggregate histogram.
Expose controls for the key levers: base delay, multiplier, max attempts, jitter mode, per-caller retry budget (e.g., 10% of request volume per Google SRE), and a "retry storm" toggle that disables budget enforcement to demonstrate how naive retries amplify load during partial outages. Preset scenarios (thundering herd, slow degradation, flapping dependency, budget exhaustion) give users a starting point without needing to hand-tune probability curves.