Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add serac-labs/serac --skill instance-performance-triagegit clone --depth 1 https://github.com/serac-labs/seracWrote 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/serac-labs/serac/instance-performance-triage)<a href="https://agentmods.dev/skills/serac-labs/serac/instance-performance-triage"><img src="https://agentmods.dev/badge/skills/serac-labs/serac/instance-performance-triage/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/serac-labs/serac/instance-performance-triage"><img src="https://agentmods.dev/badge/skills/serac-labs/serac/instance-performance-triage.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Prompt Injection · line 69 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 77 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
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 | $0.00055 | $0.05883 |
| Opus 5.5 | $0.00022 | $0.02353 |
| Sonnet 5.5 | $0.00011 | $0.01177 |
| Haiku 4.5 | $0.00006 | $0.00588 |
Grade A, and why
instance-performance-triage scanned grade A with 0 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 412 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Instance Performance Triage
Two questions arrive constantly and both have the same wrong answer: "the instance is slow" and "my
nightly job never ran". The wrong answer is to write a GlideRecord sweep and run it through
snow_execute_script. On an instance that is already struggling you have just added a transaction,
and whatever you measure includes yourself. The platform already records what you need. Read it.
This guide is about reading it correctly, because most of the failure modes here are silent — you get a plausible-looking answer that is not an answer.
Rule zero: a wrong field name returns everything
The Table API does not reject an encoded query that references a column the table does not have. It
drops the condition, logs Invalid query detected / Unknown field <x> in table <y> to syslog, and
returns rows up to sysparm_limit. Same for sysparm_fields: unknown names are simply absent from the
result, so a tool that maps them produces undefined, and a tool that counts them produces zero.
The two consequences you must hold in your head for the whole of this guide:
- A full result set does not mean your filter matched. It may mean your filter evaporated.
- An empty result set does not mean the instance is healthy. It may mean the table is not readable by the service account, or the rows were flushed, or you filtered on a value the column never holds.
So before you filter on any column named in this guide, on your instance, confirm it exists:
await snow_discover_table_fields({ table_name: "sys_trigger" })
The table-api-reads skill covers this same behaviour from the read-tool side; what follows is the
part you need in front of you while reading telemetry.
One catch: that tool queries sys_dictionary with name=<table>, which returns only the columns
defined on that table, not inherited ones. sysauto_script defines very little of its own — the
scheduling columns live on its parent sysauto — so a lookup against the child alone under-reports
what you can actually filter on. When a table extends another, query the dictionary for the whole
chain:
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 · 412 lines · 55 tokens per session scan A 6b395a20c9ad
instance-performance-triage is a skill published in the GitHub repository serac-labs/serac (78 stars, last pushed yesterday), licensed Apache-2.0. It adds 55 tokens to every session and 5,883 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
sn-safe-write
Change ServiceNow data safely — plan first, apply only after review, verify, and know how to revert. Use whenever a create, update, upsert or delete on an instance is requested.
sn-upgrade
Review ServiceNow upgrade readiness — the upgrade history, the skipped records an upgrade left unresolved, base vs customer versions of one skip, and store apps with an update. Use after a family or patch upgrade, before planning one, or when asked which skipped records to revert, keep or merge.
sn-impact
Estimate the blast radius of changing a ServiceNow table, field or script — what references it and what automation runs around it. Use before renaming, deleting or changing a field, table, script include or business rule.
sn-triage
Triage a ServiceNow incident or a failing behaviour — instance health, recent errors, what automation runs on the table, and what changed on the record. Use when something is broken, slow or erroring on an instance.
triage
Diagnose and fix broken MCP servers in mcptoon — doctor, health, per-server probes, and the common failure playbook.
sn-discover
Map a ServiceNow instance into Markdown (version and counts, custom applications, per-scope tables and artefacts) with the server's own document generator. Use when asked to discover, survey, inventory or document an instance or its custom apps.