Plain-English Runbook Generator for Operations Teams
An AI tool that watches over the shoulder of an operations or IT person (with consent), reads their scripts and cron jobs, and writes the documentation they never have time to write — in language a non-technical manager can actually use.
IT and operations leads at small companies who are the 'shell scripting priest' in their org
- Repository and server-side scanner (read-only) that ingests shell scripts, crontabs, and config files
- Auto-generated runbooks with explanation of what each script does, when it runs, and what it touches
- Risk callouts for dangerous commands (rm -rf, broad find patterns, hard-coded secrets) before handoff
- Export to Confluence, Notion, or PDF for sharing with non-technical stakeholders
The essay's 'DO NOT TOUCH – Dave 2014' crontab vignette is the lived reality the product is built to fix; documentation is the explicit gap.
Real pain validated — Scribe explicitly lists 'IT Operations' as a target function with Pro Team at $13/seat/mo, and multiple DIY tutorials (auto-runbook on GitHub, DevOpsBoys Claude runbook guide, DeciusAc GPT-for-docs) show practitioners actively trying to solve this.Scribe | Pricing ↗GitHub - vijayanandmit/auto-runbook: Automated Runbook Engine for Incident Response, DevOps, and AI-assisted Operations ↗
Adjacent tools (Scribe, Tango, Guidde) capture UI clicks, not terminal sessions or script/cron repos; existing AI-runbook tutorials are mostly K8s/incident-response focused, leaving the plain-English-from-shell-scripts niche under-served but narrow and easy for incumbents to extend into.Scribe vs Tango: Process Documentation Tool Review ↗GPT for Documentation - DeciusAc ↗
Proven per-seat pricing of $13-25/mo at SMB scale (Scribe Pro Team) and enterprise custom plans with auto-PII redaction shows ops buyers pay for doc tooling; however the target persona — a single 'shell scripting priest' at a small company — is the worst seat-count segment and is price-sensitive.Scribe | Pricing ↗Pricing and Plan Options | Tango ↗
Documentation debt and bus-factor risk are structural and grow with cloud/IaC adoption, but the specific 'over-the-shoulder screen observer' mechanic faces cultural/privacy headwinds post-COVID (employee-monitoring backlash), and the script-reading capability may be absorbed by GitHub Copilot/ChatGPT reading the repo directly.
Doable but not trivial: file/git integration with bash, cron, systemd, Terraform is heterogeneous, PII/secret redaction is a real engineering burden (Scribe gates it behind enterprise tier), and the screen-watching half requires the same consent/reputation plumbing that has hurt incumbent monitoring tools.