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 ArdurAI/ardur-agent --skill write-runbookgit clone --depth 1 https://github.com/ArdurAI/ardur-agentWrote 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/ardurai/ardur-agent/write-runbook)<a href="https://agentmods.dev/skills/ardurai/ardur-agent/write-runbook"><img src="https://agentmods.dev/badge/skills/ardurai/ardur-agent/write-runbook/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/ardurai/ardur-agent/write-runbook"><img src="https://agentmods.dev/badge/skills/ardurai/ardur-agent/write-runbook.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 | $0.00020 | $0.01143 |
| Opus 5.5 | $0.00008 | $0.00457 |
| Sonnet 5.5 | $0.00004 | $0.00229 |
| Haiku 4.5 | $0.00002 | $0.00114 |
Grade A, and why
write-runbook 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 — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Writing a runbook
A runbook turns a once-solved problem into a procedure anyone on-call can execute at 3am, half-asleep, without your context. Write it for that reader: not the expert who already knows the system, but the tired engineer who has never seen this failure before.
The test of a good runbook is simple — could someone who has never touched this service resolve the incident by following it literally? If a step requires judgement you have not written down, the runbook is not done.
When to write one
Write a runbook when a problem is recurring or likely to recur, has a known response, and is time-sensitive when it happens. A one-off that will never repeat does not need a runbook; it needs a postmortem. A problem with no known response does not need a runbook yet; it needs investigation.
Structure
Follow this order. It matches the order an on-call engineer actually needs the
information. The full skeleton with prompts is in @./template.md.
1. Symptom — how you know you are here
What does the operator observe? Be concrete and matchable:
- The exact alert name or alert text that fires.
- The dashboard panel that goes red and its threshold.
- The user-visible behavior ("checkout returns 503", "search latency > 2s").
The operator arrives holding a symptom, not a diagnosis. They must be able to match what they see against your symptom line in seconds. If two failures look identical at the symptom level, say so and point to the disambiguating check.
2. Impact — how bad, how urgent
- Who is affected and how many (all users, one region, one tenant)?
- What is degraded versus fully down?
- Is there data loss risk, or only availability?
This sets the operator's urgency and tells them whether to page for help before proceeding.
3. Diagnosis — confirm before you act
Steps to confirm this is the failure the runbook addresses, and not a look-alike. Each step is a concrete command or query with the expected output:
# Is the queue actually backed up, or is the metric stale?
$ redis-cli -h <host> LLEN jobs:pending
# Expected when healthy: < 100. If > 10000, this runbook applies.
What ships with it
1 file 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 · 121 lines · 20 tokens per session scan A 21cc6cb95d4d
write-runbook is a skill published in the GitHub repository ArdurAI/ardur-agent (2 stars, last pushed yesterday), licensed Apache-2.0. It adds 20 tokens to every session and 1,143 once invoked, about $0.0001 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 2026-08-31.
Other skills, from other repositories
worktree
Use when the user requests a Git worktree or a development task needs an isolated directory to create, find, reuse, hand off, or clean up. Exclude read-only work, non-Git repositories, and tasks that need no separate branch or directory.
member-studio
Use to create a Rovai member or revise and confirm the unsaved identity card and avatar for that creation. Exclude profile questions, edits to existing members, and character or avatar designs without roster creation.
analyze-agent-codebase
Use to analyze an agent system's architecture or mechanisms from repository evidence, including follow-up questions and analysis documents. Exclude ordinary code review, implementation, fixes, and conceptual questions that need no repository evidence.
grill-duo-with-docs
Clarify a plan or design with one fixed Thread reviewer while maintaining confirmed domain language, current specifications and version decisions. Applies to the initiator and invited reviewer during that exchange; excludes solo questions, group debates and questioning without documentation work.
review-duo
Review a defined code change with two Thread members independently checking standards, quality and requirements. Applies to the initiator and invited reviewer through the final report; excludes solo review, undefined scope and implementation-only requests.
cli-operations
Choose among Rovai messages, Missions, history and Memory; coordinate parallel member requests, multi-step operations or recovery. For one known operation, use its command help directly.