Root Cause Analysis agent for deep-dive incident investigations, hypothesis testing, and prevention follow-up. USE FOR: run post-incident root cause analysis, trace production outage to contributing factors, generate 5-why analysis report. DO NOT USE FOR: live incident triage and containment, general performance…
Use when S2 assessment data shows the work is far larger than estimated and a replanning decision is needed. USE FOR: compare actual complexity to estimate, generate retire/rewrite/replatform recommendations, and open a replanning issue. DO NOT USE FOR: routine triage or deployment execution.
Use when comparing shadow-mode and live behavior during S4 soak before cutover. USE FOR: compare error rate, latency, and divergence, flag blockers, and emit a safe go/no-go check. DO NOT USE FOR: state curation or general monitoring.
Solution architecture agent for system design, C4 diagrams, ADRs, technology selection, and cross-cutting concerns. Use when designing new systems, evaluating technology choices, or reviewing architecture for scalability and risk.
Use when issues/PRs need clustering into meaningful sprint/project groups and rollup metrics for planning or release notes. USE FOR: grouping by sprint/wave/tags, computing issue-PR-LOC-cycle-time metrics, validating group size significance, and producing release-note-ready summaries. DO NOT USE FOR: code…
Reconstructs repository history for sprint retrospectives, generating structured markdown with metrics, timelines, and actionable development tips. USE FOR: generate sprint retro document from git history, calculate sprint metrics from PR data, analyze commit timeline patterns. DO NOT USE FOR: planning next sprint…
Reliability operations specialist. USE FOR: defining SLO/SLI and error-budget policies, incident mitigation/runbook hardening, toil-reduction automation plans, resilience/capacity reviews. DO NOT USE FOR: product feature implementation, speculative architecture rewrites without an outage signal, or bypassing…
Use when analyzing takt-time JSON to compute queue length and throughput by station, rank bottlenecks, and file the weekly bottleneck report issue. USE FOR: station-level queue pressure, throughput trends, weekly issue filing, and follow-up actions. DO NOT USE FOR: dispatching work or changing replanning policy.
Use when converting manual test paths into automation candidates. Maps paths to smoke tests, regression tiers, or agent specs. ALWAYS files a GitHub Issue for every automation candidate identified.
Use when measuring dwell time at each station and flagging SLA breaches. USE FOR: compute median and p95 time at S1-S5, identify bottlenecks, and export dashboard-ready JSON. DO NOT USE FOR: dispatching work or making replanning decisions.