Documents
Get MOPs, change controls, runbooks, and incident reports written by your AI, grounded in Damira's vendor data and CCIE/ITIL templates.
Your AI writes the document. Damira makes sure it's true. Ask for a MOP, change control, runbook, or incident report in Cursor, Claude Code, or Claude Desktop. Damira supplies the required structure and the vendor-accurate facts, and your AI writes the document as markdown in your editor.
Prerequisites
- Damira plugin installed in Cursor, Claude Code, or Claude Desktop (Install)
- Optional: run
/damira:initin your project folder (Claude Code) so documents land indocuments/and every session knows your network
How It Works
- You ask your AI for a document. Include the devices, versions, and the change or incident.
- Damira supplies the template. A CCIE/ITIL checklist of the sections the document must have and what each must cover. These ship with the plugin's skills, so your AI has them in Cursor, Claude Code, and Claude Desktop.
- Damira returns the facts. Vendor docs, release notes, CVEs, an upgrade assessment, or a troubleshooting diagnosis: whatever the document needs.
- Your AI writes it. Every section filled in with exact, vendor-specific commands and expected output. It lands as a markdown file you can save, review, commit, or paste into ServiceNow or your change tool.
In Claude Code, the plugin pauses before your AI writes into documents/ if no Damira lookup succeeded in the session. A template alone isn't evidence. If a lookup failed, your AI should say so rather than fill the gaps.
Document Types
| Document | Required sections |
|---|---|
| Method of Procedure (MOP) | Purpose & scope, change description, prerequisites, pre-checks, step-by-step procedure with expected output, post-checks, rollback with trigger criteria, communication plan, sign-off |
| Change Control / RFC | RFC identification, description & business justification, probability × impact risk matrix, implementation plan, backout plan, test & validation plan, CAB checklist, post-implementation review |
| Troubleshooting Runbook | Trigger conditions, initial triage commands, diagnostic decision tree, resolution per root cause, escalation matrix, known issues & workarounds, related runbooks |
| Incident Report / Post-Mortem | Timeline, impact assessment, root cause analysis (5 Whys), corrective actions, preventive actions with owners, lessons learned, follow-up items |
The templates enforce the quality bar a CAB expects: every step has an expected output, rollback has specific trigger criteria, and the risk rating uses a matrix rather than "low risk".
Examples
Real prompts you can copy-paste:
What a MOP Looks Like
An excerpt of what your AI writes, in markdown:
Also Available As
Every document your AI writes starts as markdown, but you're not limited to that:
| Output | From | What you get |
|---|---|---|
| Word (.docx) | Any MOP or change control | After saving the markdown, ask for a Word copy. Damira converts it, writes an .html preview alongside it, and the markdown stays the source of truth — edit it and reconvert rather than hand-editing the .docx |
| Excel workbook (.xlsx) | Migrations, platform migrations, security audits, change controls, incident timelines | Ask for a spreadsheet, workbook, or tracker version instead of (or alongside) the markdown. Damira grounds every command cell against saved evidence and marks anything it can't verify |
| Topology diagram | Device configs plus saved show cdp/lldp neighbors detail output | An SVG for the document, a self-contained HTML preview, Mermaid (paste into a wiki or ticket), D2, and optionally an editable PowerPoint slide for a deck or CAB |
The workbook and diagram outputs are their own skills with their own review steps — see Automation & Generators for how the grounding and validation work.
Sharing a preview outside your own machine? Ask for the redacted copy — Damira tokenizes hostnames, IPs, AS numbers, and subnets before anything is shared, and shows you the redaction count to confirm.
Tips
- Be specific. Device names, platforms, exact versions, IPs, and the maintenance window mean fewer follow-up questions and fewer gaps.
- Research first, then write. For upgrades, get the upgrade plan or research in the same session, then ask for the MOP. Your AI already has the facts in context.
- Troubleshoot, then document. After a troubleshooting session resolves an issue, ask for the incident report while the timeline and commands are still in the conversation.
- Edit like any file. The document is plain markdown. Ask your AI to tighten the rollback section or change the risk rating, or edit it yourself.
- Watch for UNVERIFIED. Anything your AI couldn't ground in Damira's data is marked for you to confirm against vendor documentation before the window.
Related
- Upgrade Planning: the assessment behind upgrade MOPs and change controls
- Troubleshooting: diagnosis that feeds incident reports and runbooks
- Automation & Generators: Terraform, automation scripts, CI pipelines, and tests — validated locally before they go anywhere near production
- MCP Server: the full tool reference