Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/StarRocks/starrocks-debug-skillsnpx agentmods add skills/starrocks/starrocks-debug-skills/shared-dataWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/starrocks/starrocks-debug-skills/shared-data)<a href="https://agentmods.dev/skills/starrocks/starrocks-debug-skills/shared-data"><img src="https://agentmods.dev/badge/skills/starrocks/starrocks-debug-skills/shared-data/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/starrocks/starrocks-debug-skills/shared-data"><img src="https://agentmods.dev/badge/skills/starrocks/starrocks-debug-skills/shared-data.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00077 | $0.07284 |
| Opus 5.5 | $0.00031 | $0.02914 |
| Sonnet 5.5 | $0.00015 | $0.01457 |
| Haiku 4.5 | $0.00008 | $0.00728 |
Grade C, and why
shared-data scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf <datacache_disk_path>/* Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -s "http://$be:8040/metrics" | grep "diskcache_disk_avail_bytes" How it starts
The opening of the file, as written. The whole thing — 769 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Shared-Data Troubleshooting
Investigation guide for shared-data architecture issues: DataCache (StarCache) corruption and autoscaling regressions, S3 rate limiting and multipart upload failures, FE leader switch bugs, and compaction score management.
Five root causes account for the vast majority of cases:
- Cause A — DataCache autoscaling regression (eviction triggers S3 reads, compaction slows)
- Cause B — DataCache corruption (checksum mismatch causing publish or query failures)
- Cause C — S3 rate limiting (503 errors or multipart upload failures)
- Cause D — FE leader switch issues (EditLog gap causing transaction replay failure)
- Cause E — Cached tablet metadata inconsistency (stale cache version mismatch)
Metric Taxonomy — Read This First
Before using any metrics, understand the key observability sources:
DataCache Metrics
| Metric / Source | Meaning |
|---|---|
starrocks_be_diskcache_disk_avail_bytes |
Remaining free bytes on the DataCache disk volume — when near zero, autoscaler evicts data |
| DataCache hit rate | Ratio of cache hits to total accesses — a drop indicates eviction or corruption; visible in Grafana |
be.WARNING — star cache autoscaling |
Log messages when DataCache triggers autoscaling eviction |
be.WARNING — checksum mismatch / corrupted compressed block |
Cache block integrity failure — data on disk does not match expected checksum |
Compaction Metrics (Shared-Data)
| Metric / Source | Meaning |
|---|---|
information_schema.be_cloud_native_compactions |
Cloud-native compaction history: tablet, timing, size, status — primary compaction audit table for shared-data |
lake_publish_tablet_version_queuing_count |
Number of tablet publish tasks queued — high value means publish is falling behind |
| BE log — compaction score warnings | Score thresholds exceeded; indicates compaction cannot keep up with ingestion |
S3 / Object Storage Metrics
| Metric / Source | Meaning |
|---|---|
| fslib write IO metrics — p99 latency | S3 PUT/GET tail latency — spikes indicate throttling or network degradation |
503 rate limit errors |
S3 request rate exceeded — visible in BE log |
| S3 multipart upload failure log | Partial uploads that leave .dat / .sst orphan fragments |
lake_flush_thread_num_per_store pool depth |
Flush thread pool queue depth — saturation means S3 writes are blocking flush |
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 769 lines · 77 tokens per session scan C 547f1f7086f2
shared-data is a skill published in the GitHub repository StarRocks/starrocks-debug-skills (77 stars, last pushed 5mo ago), licensed Apache-2.0. It adds 77 tokens to every session and 7,284 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
application-integration-troubleshooting
Diagnoses failed, stuck, or silent Google Cloud Application Integration runs and Integration Connectors tasks using the REST API, Cloud Logging, and Cloud Monitoring. Use when an integration fails, returns HTTP 200 without doing the expected work, hangs in a suspended or retry state, or hits IAM, run-as service…
gke-workload-troubleshooting
Diagnoses GKE workload failures (CrashLoopBackOff, OOMKilled, ImagePullBackOff, Pending, etc.) via logs and events. Use when pods fail to start or crash repeatedly. Don't use for GKE cluster infrastructure provisioning, node pool creation, or non-Kubernetes Google Cloud services.
gke-workload-scaling-troubleshooting
Diagnoses GKE HorizontalPodAutoscaler (HPA) failures — metrics showing as , FailedGetResourceMetric / FailedGetScale / FailedComputeMetricsReplicas events, missing Pod resource requests, custom/external metrics-pipeline breakage (FailedGetExternalMetric / FailedGetCustomMetric, unavailable metrics adapter…
gke-ai-troubleshooting-tpu-mxla-hang
Diagnose GKE Cloud TPU multi-slice training hangs (Megascale HANGDETECTED logs and hang monitored events) using the ML Diagnostics Megascale XLA (MXLA) Hang Analyzer (gcloud alpha mldiagnostics monitored-events) and 1-minute Cloud Monitoring multi-slice latency metrics (kubernetes.io/container/multislice/).…
gke-ai-troubleshooting-tpu-performance-degradation
Diagnose GKE Cloud TPU training throughput drops and step-time regressions (15%+ TPU duty-cycle drop) using ML Diagnostics Workload Monitoring (gcloud alpha mldiagnostics monitored-events / hypercomputecluster.googleapis.com/v1alpha) and 1-minute Cloud Monitoring system metrics (kubernetes.io/node/accelerator/).…
gke-ai-troubleshooting-jobset-interruption
Diagnoses GKE JobSet interruptions, restarts, and preemptions for AI/ML training workloads autonomously. Use when troubleshooting JobSet restart loops, spot VM preemptions, node readiness failures, host VM issues, or coordinator worker crashes. Don't use for general GKE cluster creation, basic workload deployment, or…