agentsclimarketplace

Debug & triage

5,150 rows, by stacks then stars

1,113 of these skill files have been read, by 1,020 distinct authors, and what they tell an agent counted. What Debug & triage authors agree on, and what they forbid

  • Logging architect

    AtulPurohit/Antigravity-Awesome-Skills/plugins/devops-cloud-infra/skills/logging-architect Skill

    no license5★ repo

    Professional Logging Architect skill. Build performant, secure, and scalable backend logic and RESTful or GraphQL APIs.

  • Debugging expert

    AtulPurohit/Antigravity-Awesome-Skills/plugins/dx-qa-automation/skills/debugging-expert Skill

    no license5★ repo

    Professional Debugging Expert skill. Build performant, secure, and scalable backend logic and RESTful or GraphQL APIs.

  • Debugging expert

    AtulPurohit/Antigravity-Awesome-Skills/skills/debugging-expert Skill

    no license5★ repo

    Professional Debugging Expert skill. Build performant, secure, and scalable backend logic and RESTful or GraphQL APIs.

  • Logging architect

    AtulPurohit/Antigravity-Awesome-Skills/skills/logging-architect Skill

    no license5★ repo

    Professional Logging Architect skill. Build performant, secure, and scalable backend logic and RESTful or GraphQL APIs.

  • Cds debug

    Enakoneschniy/claude-dev-stack/vendor/cds-workflow/skills/cds-debug Skill

    5★ repo

    Systematic debugging with persistent state across context resets

  • Gsd debug

    Enakoneschniy/claude-dev-stack/vendor/cds-workflow/skills/gsd-debug Skill

    5★ repo

    [DEPRECATED] Use /cds-debug instead

  • Logs

    cyrus-cai/claude-cobrain/plugin/skills/logs Skill

    5★ repo

    View cobrain daemon logs.

  • Log monitoring playbook

    maxwellokumu/okaudit-claude-skills/log-monitoring/log-monitoring-playbook Skill

    no license5★ repo

    Guide Claude through a structured logging and monitoring audit covering log inventory, baseline analysis, anomaly review, incident investigation, and reporting.

  • Frontend bugfix debugger

    ytvee-dev/webdev-agent-kit/skills/frontend-bugfix-debugger Skill

    5★ repo

    Diagnose and fix frontend defects from evidence, including unclear UI/runtime errors, broken routes, styling regressions, and hydration or client issues. Reproduce before editing; exclude planned refactors and backend-only debugging.

  • Debug

    DongDuong2001/pudo-code-system/skills/debug Skill

    5★ repo

    Modular, scalable, automated development intelligence.

  • Crash analysis

    DongDuong2001/pudo-code-system/skills/debug/crash-analysis Skill

    5★ repo

    Analyzing core dumps, stack traces, and unhandled exceptions

  • Memory

    DongDuong2001/pudo-code-system/skills/debug/memory Skill

    5★ repo

    Debugging memory leaks, out-of-memory errors, and garbage collection

  • Network

    DongDuong2001/pudo-code-system/skills/debug/network Skill

    5★ repo

    Investigating API failures, DNS issues, and connection timeouts

  • Performance

    DongDuong2001/pudo-code-system/skills/debug/performance Skill

    5★ repo

    Diagnosing and resolving performance bottlenecks and latency issues

  • Os memory manager

    richfrem/agent-plugins-skills/plugins/agent-agentic-os/skills/os-memory-manager Skill

    5★ repo

    Trigger with "remember this", "update memory", "what should we record from this session", "capture learnings", "write a session log", or when closing a session. Guides agents on managing memory hygiene across sessions, deciding what to write to dated memory logs, what to promote to long-term memory.md, and when to archive. <example> User: I'm done for the day, can you write up a session log? Agent: <Bash> python context/kernel.py emit_event --agent os-memory-manager --type intent --action promote_memory python context/kernel.py state_update active_agent os-memory-manager </Bash> </example> <example> User: That's all, logging off now. Agent: <Bash> python context/kernel.py acquire_lock memory </Bash> </example> <example> User: How does the memory system work? Agent: <Read> ./references/architecture/context-folder-patterns.md </Read> </example>

    The same file is in 1 other repository

  • Github issue agent

    richfrem/agent-plugins-skills/plugins/dev-utils/skills/github-issue-agent Skill

    5★ repo

    Agent skill for safe, dry-run-first, deduplicated, root-cause-consolidated, evidence-validated, and secret-redacted logging of repository execution friction into GitHub Issues. USE ONLY for durable repository bugs, execution friction (T1-T3), map debt, and architectural improvements. DO NOT USE for temporary intra-session checklists (use task-agent instead).

  • Github issue backlog agent

    richfrem/agent-plugins-skills/plugins/dev-utils/skills/github-issue-backlog-agent Skill

    5★ repo

    Bridge skill for escalating ephemeral local task scratchpad items (`tasks/*.md`) into durable, taxonomy-validated, evidence-rich GitHub Issues. USE ONLY when promoting a single-session local task into durable repository backlog. DO NOT USE for managing local kanban boards (use `task-agent` instead) or directly querying/commenting on issues (use `github-issue-agent` instead).

  • Github issue prioritizer

    richfrem/agent-plugins-skills/plugins/dev-utils/skills/github-issue-prioritizer Skill

    5★ repo

    Automatically ranks GitHub issues (P0-P3) based on friction tier, frequency, and blockages, synchronizing priority labels and GitHub Projects v2 custom fields.

  • Csat root cause

    SkillMedev/skills/skills/csat-root-cause Skill

    5★ repo

    Analyzes CSAT and NPS verbatims to surface root-cause themes and prioritize fixes by frequency and impact. Use when interpreting customer satisfaction data to inform product or process decisions.

    The same file is in 1 other repository

  • Email triage

    SkillMedev/skills/skills/email-triage Skill

    5★ repo

    Triages a batch of emails into delete, do-now, delegate, or defer decisions using the 2-minute rule, returning drafted replies, delegation forwards, and dated tasks. Use when someone says "help me get through my inbox", "triage these emails", "I have 200 unread messages, what needs a reply", or "turn this pile of email into a to-do list". Do NOT use for designing the ongoing system that keeps an inbox at zero - filters, folder setup, unsubscribe habits, daily routines - use inbox-zero instead.

  • Error handling

    SkillMedev/skills/skills/error-handling Skill

    5★ repo

    Designs typed, observable, recoverable error handling - an explicit taxonomy of retryable vs terminal failures, Result types at boundaries, single-point logging, and retry policies with real backoff numbers. Use when someone asks "should I retry this error", "how do I structure error types", "why are my logs full of duplicate stack traces", "where do I put the try/catch", or is designing failure paths for an API, job, or client. Do NOT use for building the circuit breaker component itself - use circuit-breaker-builder instead; do NOT use for handling 429s against third-party APIs - use rate-limit-handler instead; do NOT use for writing user-facing error copy - use error-message-writer instead.

    The same file is in 1 other repository

  • Error message writer

    SkillMedev/skills/skills/error-message-writer Skill

    5★ repo

    Rewrites error messages to be specific, actionable, and blame-free - what happened, why it helps to know, and what to do next - with tone calibrated to severity and developer-facing detail kept separate from the user-facing text. Use when someone asks "rewrite this error message", "our app just says something went wrong", "make this validation message less hostile", or is auditing error states before launch. Do NOT use for reviewing all in-product copy beyond errors - use ux-writing-audit instead; for the copy users see during first-run setup, use onboarding-copy.

  • Pii scrubber

    SkillMedev/skills/skills/pii-scrubber Skill

    5★ repo

    Detects and redacts personally identifiable information from text, logs, datasets, and LLM prompts using layered pattern, checksum, and NER detection, then picks the right redaction mode for each downstream use. Use when someone asks "scrub PII from these logs", "redact personal data before sending it to the LLM", "is this dataset safe to share", or is wiring up logging, analytics, or third-party exports that might carry user data. Do NOT use for leaked API keys, tokens, or credentials - use secrets-hygiene instead.

    The same file is in 1 other repository

  • Replication lag debugger

    SkillMedev/skills/skills/replication-lag-debugger Skill

    5★ repo

    Diagnoses read-replica lag and the stale-read bugs it causes, then applies read-after-write consistency strategies to fix them. Use when someone says "users see stale data right after saving", "my read replica is lagging behind the primary", "data written a second ago is missing from the next request", when replica lag monitoring alerts or grows after a backfill, or when deciding which reads must hit the primary versus a replica.

    The same file is in 1 other repository

  • Sev triage

    SkillMedev/skills/skills/sev-triage Skill

    5★ repo

    Classifies incident severity (SEV1-4) using impact, scope, and urgency signals and decides who to page. Use when an alert fires or a report comes in and a severity call must be made quickly.

    The same file is in 1 other repository

  • Vulnerability triage

    SkillMedev/skills/skills/vulnerability-triage Skill

    5★ repo

    Prioritizes vulnerability findings from scanners, pentest reports, and bug bounty submissions by real-world exploitability rather than raw CVSS, assigning internal severity tiers with fix SLAs. Use when someone asks "which of these CVEs do we fix first", "triage this scanner report", "is this CVSS 9.8 actually critical for us", or is facing a wall of security findings and needs an actionable queue. Do NOT use for enumerating threats in a design that has not shipped - use threat-model-stride instead; for an active exploitation incident in production, use sev-triage; for reviewing the code itself, use secure-code-review.

    The same file is in 1 other repository

  • Intake triage

    HsuanYuLee/polaris/.claude/skills/intake-triage Skill

    5★ repo

    批次收單排工:分析 PM 開出的一批 ticket,評估優先序,告訴 PM 哪些先做、哪些後做、哪些規格不足需補。 產出 JIRA label + comment + Slack 摘要。 觸發:「收單」、「排工」、「intake」、「這批單幫我看」、「PM 開了一堆單」、「幫我排優先」、 「intake-triage」、「triage these tickets」、「prioritize this batch」。 當使用者提供一批 ticket key 並要求排優先序時使用此 skill,不要與 /my-triage(個人每日盤點)混淆。

  • My triage

    HsuanYuLee/polaris/.claude/skills/my-triage Skill

    5★ repo

    個人工作盤點與 zero-input next router;列出 assigned Epics/Bugs/Tasks,整合 cross-session resume signals,協助決定下一步。

  • Bug fix

    ramboz/jig/hosts/claude/skills/bug-fix Skill

    5★ repo

    Drive the teeth-gated lifecycle for reported defects: diagnose root cause, prove it, and prevent regression through REPORTED → DIAGNOSING → ROOT_CAUSED → FIXING → REVIEWED → DONE, with VERIFIED, ESCALATED, and RESOLVED_ON_MAIN paths where needed. Auto-fires on fix this bug, debug this, root-cause this, this regressed, this broke again, why is this failing, diagnose before fixing, or investigate this failure. Two modes: `diagnose` stops at ROOT_CAUSED; `diagnose_and_fix` runs through DONE. Uses a durable bug record, multiple hypotheses, a fresh-main recheck, and a witnessed red→green test. Do not use for spec-shaped work (use `spec-workflow`) — including a pure visual design-fidelity gap against an agreed mockup, which is spec-shaped, not a bug (see "Design-fidelity triage" below) — or trivial one-liners (use `tdd-loop`).

  • Bug fix

    ramboz/jig/hosts/codex/plugins/jig/skills/bug-fix Skill

    5★ repo

    Drive the teeth-gated lifecycle for reported defects: diagnose root cause, prove it, and prevent regression through REPORTED → DIAGNOSING → ROOT_CAUSED → FIXING → REVIEWED → DONE, with VERIFIED, ESCALATED, and RESOLVED_ON_MAIN paths where needed. Auto-fires on fix this bug, debug this, root-cause this, this regressed, this broke again, why is this failing, diagnose before fixing, or investigate this failure. Two modes: `diagnose` stops at ROOT_CAUSED; `diagnose_and_fix` runs through DONE. Uses a durable bug record, multiple hypotheses, a fresh-main recheck, and a witnessed red→green test. Do not use for spec-shaped work (use `spec-workflow`) or trivial one-liners (use `tdd-loop`).

  • Bug fix

    ramboz/jig/skills/bug-fix Skill

    5★ repo

    Drive the teeth-gated lifecycle for reported defects: diagnose root cause, prove it, and prevent regression through REPORTED → DIAGNOSING → ROOT_CAUSED → FIXING → REVIEWED → DONE, with VERIFIED, ESCALATED, and RESOLVED_ON_MAIN paths where needed. Auto-fires on fix this bug, debug this, root-cause this, this regressed, this broke again, why is this failing, diagnose before fixing, or investigate this failure. Two modes: `diagnose` stops at ROOT_CAUSED; `diagnose_and_fix` runs through DONE. Uses a durable bug record, multiple hypotheses, a fresh-main recheck, and a witnessed red→green test. Do not use for spec-shaped work (use `spec-workflow`) — including a pure visual design-fidelity gap against an agreed mockup, which is spec-shaped, not a bug (see "Design-fidelity triage" below) — or trivial one-liners (use `tdd-loop`).

  • Diagnose

    shalintripathi/organic-os/plugin/skills/diagnose Skill

    5★ repo

    Use to collect a bug-report diagnostic - "diagnose", "collect debug info", /organic-os:diagnose. Prints a report the user can paste into an issue; never transmits it anywhere.

  • Hyper debug

    zeikar/hyperclaude/skills/hyper-debug Skill

    5★ repo

    Use when encountering a test failure, exception, regression, or unexpected behavior — before proposing fixes. Also when the user invokes /hyperclaude:hyper-debug. Forms hypotheses systematically instead of guess-and-check thrashing.

  • Logging operator

    air-gapped/skills/.claude/skills/logging-operator Skill

    5★ repo

    Configure and operate the kube-logging logging-operator (formerly Banzai Cloud) on Kubernetes — the CRD-driven log pipeline: Fluent Bit collector → fluentd or syslog-ng aggregator → outputs. Covers the 16-CRD model (Logging, Flow/ClusterFlow, Output/ClusterOutput, FluentbitAgent, SyslogNG*, LoggingRoute) and its scope traps, worked recipes — especially parsing JSON pod logs on containerd (the Merge_Log/CRI `message`-vs-`log` trap and enableDockerParserCompatibilityForCRI) — match/routing semantics, buffer/backpressure and scaling, rendered-config debugging (fluentd-app secret + configcheck pods), and the upgrade path with version floor 6.7.0 (CVE-2026-54680 config-injection RCE).

  • Rancher logging exit

    air-gapped/skills/.claude/skills/rancher-logging-exit Skill

    5★ repo

    Migrate off the Rancher-bundled `rancher-logging` chart (cattle-logging-system, rancher/mirrored-kube-logging-* images) to the upstream kube-logging logging-operator ≥6.7.0 — air-gap-first. Rancher 2.11 through 2.15-dev all bundle a frozen operator 4.10.0 that is inside the affected range of CVE-2026-54680 (CVSS 9.9 config-injection RCE, no SUSE fix) — so the exit is security-urgent. Covers the maintainer-endorsed helm-release-secret strategy (near-zero gap; NOT `helm uninstall rancher-logging-crd`, which cascade-deletes every CR and the data plane), CR compatibility 4.10→6.7 (silent field pruning), server-side CRD apply (828KB CRDs), buffer-PVC preservation, air-gap image/chart mirroring, rollback, and stale-CRD debris cleanup.

  • Triage

    air-gapped/skills/.claude/skills/triage Skill

    5★ repo

    [3/4 defending-code] Triage a batch of raw security findings. Third step of the find-and-fix loop (/threat-model -> /vuln-scan -> /triage -> /patch); consumes /vuln-scan's VULN-FINDINGS.json, but accepts any scanner output, so it also stands alone on a third-party backlog. Verify each is real, collapse duplicates, re-rank by impact-on-asset x exploitability, and tag with an owner. Takes a directory or file of scanner output and writes TRIAGE.json + TRIAGE.md sorted by what actually needs engineering attention. Use when asked to "triage findings", "validate scanner output", "prioritize vulns", or "review the backlog". Runs interactively by default; pass --auto to skip the interview.

  • Vllm observability

    air-gapped/skills/.claude/skills/vllm-observability Skill

    5★ repo

    Observe production vLLM — `/metrics` Prometheus surface (V1 engine), SLO-driven alerting on TTFT/ITL/queue/KV/preemption/aborts/corrupted-logits, shipping Grafana dashboards in `examples/observability/`, OTLP tracing with `--otlp-traces-endpoint` and `--collect-detailed-traces={model,worker,all}`, diagnostic rules to triage from /metrics alone — queue-grows + TPOT-stable means capacity, queue-stable + TPOT-grows means context/model, DCGM `SM_OCCUPANCY` is the real GPU-saturation signal not `GPU_UTIL`. V1 metric names (kv_cache_usage_perc), gpu_→kv_ rename saga, DCGM-exporter pairing, dashboard-lying pitfalls.

  • Vllm tool parsers

    air-gapped/skills/.claude/skills/vllm-tool-parsers Skill

    5★ repo

    vLLM tool-calling operator reference — picking `--tool-call-parser` per model family, writing custom parsers via `--tool-parser-plugin`, navigating vLLM source + GitHub tracker to debug any specific tool-call question. Pointer map, not source paraphrase. All 40+ built-in parsers (JSON-sentinel, pythonic, XML, harmony grammars), CLI contract, `prev_tool_call_arr`/`streamed_args_for_tool` flush invariants, diagnostic playbook (isolate template-vs-parser via raw `/v1/completions` + unicodedata codepoint decode).

  • Power bi performance troubleshooting

    fabioc-aloha/Alex_Skill_Mall/plugins/data-analytics/power-bi-performance-troubleshooting Skill

    5★ repo

    Systematic Power BI performance troubleshooting prompt for identifying, diagnosing, and resolving performance issues in Power BI models, reports, and queries.

  • Debugging patterns

    fabioc-aloha/Alex_Skill_Mall/plugins/reasoning-metacognition/debugging-patterns Skill

    5★ repo

    Systematic problem-solving and error analysis.

  • Hypothesis driven debugging

    fabioc-aloha/Alex_Skill_Mall/plugins/reasoning-metacognition/hypothesis-driven-debugging Skill

    5★ repo

    Investigate build failures, test errors, runtime crashes, or unexpected behavior through systematic minimal reproduction, multi-hypothesis testing, and evidence-based verification. Use when asked to "debug", "investigate a failure", "find root cause", "troubleshoot", "fix a failing test", or "diagnose an issue". Always re-run builds and tests after changes.

  • Root cause analysis

    fabioc-aloha/Alex_Skill_Mall/plugins/reasoning-metacognition/root-cause-analysis Skill

    5★ repo

    Find the true source, not symptoms — systematic debugging from observation to permanent fix

  • Rubber duck debugging

    fabioc-aloha/Alex_Skill_Mall/plugins/reasoning-metacognition/rubber-duck-debugging Skill

    5★ repo

    Be a thinking partner. The answer often emerges when explaining the problem.

  • Error message sanitization

    fabioc-aloha/Alex_Skill_Mall/plugins/security-privacy/error-message-sanitization Skill

    5★ repo

    Raw error messages leak internal information to users:

  • Feedback triage

    fabioc-aloha/Alex_Skill_Mall/plugins/supervisor-fleet/feedback-triage Skill

    5★ repo

    Categorize, deduplicate, and route fleet feedback to Edition fix, Mall change, or AlexMaster escalation

  • Unity debug profiler

    tea-x-random/unity-game-skills/skills/unity-debug-profiler Skill

    5★ repo

    Debug and profile casual iOS games in Unity through the unity-mcp-bridge. Use for compile errors, console errors, NullReferenceException, MissingReferenceException, missing/broken references, pink/magenta materials, blank or black scenes, won't enter Play Mode, domain-reload issues, MCP connection drops, and for performance profiling: frame rate, draw calls/batches/SetPass, GC allocations, memory and texture memory, mobile thermal throttling, IL2CPP/AOT managed-code stripping, and slow builds. Evidence-driven: read the actual console and profiler, verify Unity 6 APIs via MCP, and re-run to confirm fixes.

  • Awesome logging standards

    khasky/awesome-agent-skills/skills/awesome-logging-standards Skill

    5★ repo

    Skills for AI coding agents: code review, debugging, security audits, refactoring, cleaning up AI-written code and text, and more.

  • Awesome root cause

    khasky/awesome-agent-skills/skills/awesome-root-cause Skill

    5★ repo

    Skills for AI coding agents: code review, debugging, security audits, refactoring, cleaning up AI-written code and text, and more.

  • Ai ecommerce review root cause analysis

    skillsaiagent/aiskills/skills/ai-ecommerce-review-root-cause-analysis Skill

    no license5★ repo

    业务诊断助手适合运营、市场营销、销售、产品在用户提出“这件事该怎么做”这类问题,需要快速拆解目标、判断重点并形成可执行结果时使用,帮助基于输入材料生成摘要、诊断结论、行动建议和可复用交付物。

  • Ai equipment failure rca draft

    skillsaiagent/aiskills/skills/ai-equipment-failure-rca-draft Skill

    no license5★ repo

    视觉质量诊断助手适合运营、管理者、manufacturing、RCA在用户提出“这张图够发布吗”这类问题,需要快速拆解目标、判断重点并形成可执行结果时使用,帮助基于输入材料生成摘要、诊断结论、行动建议和可复用交付物。

  • Bureau cyber

    rikitrader/glaw/bureau-cyber Skill

    no license5★ repo

    GLAW Investigations Bureau — the Cyber Intelligence Agent. The digital-evidence specialist: triages malware artifacts (read-only), plans defensible forensic imaging and hashing (FRE 902(14)), runs OSINT and dark-web intelligence ANALYSIS, hunts threats, analyzes network logs, builds attribution, and frames incident response. Works only from lawfully-obtained data — never intrudes a live system. Use for: 'malware analysis', 'digital forensics plan', 'forensic imaging', 'hash verification', 'dark web intelligence', 'threat hunting', 'network log analysis', 'attribution', 'incident response', 'cyber investigation'.

  • Konnect gateway triage

    Kong/ai-marketplace/plugins/kong-konnect/skills/konnect-gateway-triage Skill

    5★ repo

    Use when triaging Konnect Gateway control-plane and data-plane failures such as disconnected planes, missing rollout, or wrong environment slices; separate attachment, network, live-state drift, and traffic-path failures before handing off fixes.

  • Konnect observability triage

    Kong/ai-marketplace/plugins/kong-konnect/skills/konnect-observability-triage Skill

    5★ repo

    Diagnose missing, partial, delayed, or mis-scoped Konnect observability data across analytics, Explorer, and Debugger. Use when the question is dataset, scope, or telemetry visibility, not pure gateway health or operator-access troubleshooting.

  • Audit exception handling

    danygiguere/audit-skills/.agents/skills/audit-exception-handling Skill

    5★ repo

    Exception and error-handling audit checklist. Use when reviewing try/catch blocks, error propagation, empty or overly broad catches, rethrowing, cleanup/finally paths, async error handling, or whether errors map to the right HTTP status (404 vs 403, 401, 422, 409).

  • Optimize cpu workloads

    zystem-io/zymtrace-skills/zymtrace/skills/optimize-cpu-workloads Skill

    5★ repo

    Analyze CPU performance with zymtrace on CPU-only deployments (no GPUs) — rank the top CPU consumers (which process/function/container/host/pod is hottest), find what's eating CPU, pick the biggest-ROI app to optimize, or read a CPU flamegraph. The zymtrace MCP pulls the data (rankings, host metrics, flamegraphs); YOU analyze — name the hot stacks, identify the pattern (lock contention, allocation churn, serialization, GC, syscall-heavy), then recommend AND apply the fix in the user's source (ask for the path if it isn't local) and close with a follow-up. Scope to code the user controls (kube-proxy, k3s, systemd are context, not action items). Also for "investigate/troubleshoot/diagnose/look into" CPU or host performance. **DEFAULT when the user names only an entity with no accelerator hint** ("investigate my `payments` deployment", "what's going on with this container/pod/host", "why is this service slow") — CPU is the universal baseline (every workload has a CPU profile; GPU is the special case); pull metrics first and hand off to optimize-gpu-workloads if they show real GPU activity. Anything mentioning GPU, CUDA, an accelerator, inference, vLLM, SGLang, or Triton goes to optimize-gpu-workloads instead. Trigger phrases: "use zymtrace to analyze CPU profiles", "investigate/troubleshoot/diagnose my CPU usage", "investigate my deployment/container/pod/host/service", "what's going on with <service>", "why is this service slow", "what's eating my CPU", "which process/function uses the most CPU core", "find the most expensive functions/containers/app/service", "rank my apps by CPU", "biggest ROI optimization", "what should I optimize first", "where's the CPU bottleneck", "why is this service CPU-bound", "analyze the CPU flamegraph", "top CPU consumers".

  • Optimize gpu workloads

    zystem-io/zymtrace-skills/zymtrace/skills/optimize-gpu-workloads Skill

    5★ repo

    Analyze GPU performance with zymtrace — a GPU/CUDA workload, a GPU flamegraph, the hot kernel, why the GPU isn't saturated, or an inference server (vLLM, SGLang, NVIDIA Dynamo-Triton, TensorRT-LLM) or training/fine-tuning job. The zymtrace MCP pulls the data (GPU metrics, flamegraphs); YOU analyze. ALWAYS cross-check the CPU side with the same filter — a GPU workload's bottleneck often hides host-side (tokenizer, DataLoader, Python overhead, sync points) — so pull GPU metrics, the GPU flamegraph, AND the matching CPU flamegraph. Scope to code the user controls; recommend AND apply the fix in the user's source (ask for the path if it isn't local), then close with a follow-up. ANY request mentioning GPU, CUDA, an accelerator, inference, an inference server, or training/fine-tuning routes HERE — including "investigate/troubleshoot/diagnose/look into/what's wrong with" phrasings. (Outright broken rather than slow — no GPU profiles, NVML missing, pods crash-looping — use troubleshoot-zymtrace-profiler.) CPU-only workload with no GPUs → optimize-cpu-workloads. Trigger phrases: "analyze/investigate/troubleshoot/diagnose my GPU workload", "investigate my inference/training/fine-tuning job", "look into my vLLM/SGLang/Triton/TensorRT-LLM workload", "what's wrong with my GPU job", "where's the bottleneck in vllm/sglang/triton", "find the hot kernel", "GPU isn't saturated", "GPU utilization is low", "why is my GPU idle", "why is inference slow", "analyze the GPU flamegraph", "which kernel uses the most GPU", "investigate my CUDA workload", "profile my training job".

  • Troubleshoot zymtrace backend

    zystem-io/zymtrace-skills/zymtrace/skills/troubleshoot-zymtrace-backend Skill

    5★ repo

    Use when a deployed zymtrace backend isn't working as expected — no data appearing in the UI, profiles not arriving, ingest errors, ClickHouse storage full, license / auth failures, slow queries. Walks symptom → diagnosis → fix. Routes between backend (ingest, ClickHouse) and profiler (DaemonSet, CUDA injection) checks. Trigger phrases: "zymtrace not working", "no data in zymtrace UI", "zymtrace UI is empty", "no profiles appearing", "profiles not showing up", "ingest is failing", "clickhouse disk full", "license error in zymtrace", "license expired", "zymtrace queries are slow", "zymtrace broken after upgrade", "fix zymtrace", "diagnose zymtrace".

  • Troubleshoot zymtrace profiler

    zystem-io/zymtrace-skills/zymtrace/skills/troubleshoot-zymtrace-profiler Skill

    5★ repo

    Use when the zymtrace profiler agent is misbehaving — pods crash-looping, image pull errors, OOMKilled, CPU profiles working but no GPU profiles, NVML library not found, PC sampling not producing SASS-level data, or license errors on the profiler side. Walks symptom → diagnosis → fix. Focused on the agent itself; for "no data anywhere" use troubleshoot-zymtrace-backend. Trigger phrases: "profiler not working", "profiler pods crashing", "profiler CrashLoopBackOff", "profiler ImagePullBackOff", "agent OOMKilled", "no GPU profiles", "CPU profiles work but GPU doesn't", "NVML library not found", "PC sampling not working", "profiler license rejected", "agent restart cycle", "zymtrace agent unhealthy", "fix the profiler".

  • Threejs debugging

    linegel/threejs-complete-set-of-skill/skills/threejs-debugging Skill

    5★ repo

    Diagnose unexpected Three.js WebGPU/TSL runtime, rendering, API, asset, or version behavior. Use for a concrete failure, a suspected upstream regression or known issue, or a choice among an application fix, released upgrade, bounded workaround, upstream report, and blocker.

  • Debug coach

    Far-200/think-before-code/skills/debug-coach Skill

    5★ repo

    Socratic Agent Skills for learning problem-solving and software-engineering judgment — one question at a time.