Retry strategy visualization pattern
Skill kjuhwa/skills-hub/skills/design/retry-strategy-visualization-pattern
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-visualization-patternAssembled 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
Visualize retry attempts as timeline tracks with backoff intervals, jitter bands, and terminal outcome markers
SKILL.md
1.9 KB, 341 tokens by cl100k_base, as published. Nobody here has run it
retry-strategy-visualization-pattern
Render each in-flight request as a horizontal track on a time axis, with attempt markers (circles) placed at t₀, t₀+delay₁, t₀+delay₂, ... where delay_n follows the chosen backoff policy (fixed, linear, exponential, decorrelated jitter). Color-code each attempt by outcome: green for success, amber for retryable failure (429/503/timeout), red for terminal failure (4xx non-retryable), and gray for attempts suppressed by budget or circuit breaker. Overlay the computed backoff interval as a translucent band between markers so viewers can distinguish wait time from actual request latency.
For exponential-backoff strategies, draw the theoretical delay curve (base · 2^n) as a dashed reference line behind the actual attempt markers; jittered delays then visibly scatter around that curve, making the jitter distribution legible at a glance. Add a secondary panel showing aggregate metrics updated per tick: success rate, p50/p99 attempts-to-success, total retries issued, and budget consumption (retries / total_requests). A small legend should map every attempt-state color and annotate the active strategy parameters (base delay, max attempts, jitter mode, budget cap).
Keep the canvas scrolling horizontally in a fixed time window (e.g., last 30 seconds) so long-running simulations don't compress into unreadable density. Sticky the y-axis labels (request IDs) and let users hover any attempt marker to see exact delay, HTTP status, and retry-after header value. This pattern scales from single-request debugging (one track, zoomed-in) to storm analysis (hundreds of tracks, zoomed-out heatmap view).