UNPKG

aiwg

Version:

Deployment tool and support utility for AI context. Copies agents, skills, commands, rules, and behaviors into the paths each AI platform reads (Claude Code, Codex, Copilot, Cursor, Warp, OpenClaw, and 6 more) so one source of truth works across 10 platfo

aiwg.io

jmagly/aiwg

60 lines (44 loc) • 3.35 kB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
--- dod_id: dod-runbook name: Runbook Definition of Done scope: operational category: runbook version: 1.0.0 extensible: true --- # Runbook Definition of Done ## Purpose Defines when a runbook is fit for use by an on-call engineer who did not write the system. A runbook that only its author can follow is not a runbook — it is tribal knowledge with a document wrapper. Runbooks must be accurate, complete, and independently executable under incident pressure. ## Criteria ### Required - [ ] Runbook is stored in version control (not a personal wiki, not a shared drive folder, not someone's notes app) - [ ] Runbook has a clear title, service name, and last-verified date in the header - [ ] Every step is written as an imperative action with the exact command or UI action (no "run the deployment script" — write the command) - [ ] All commands include the expected output or success indicator so the operator knows the step succeeded - [ ] Prerequisites listed: access rights, tools, and environment variables required before starting - [ ] Runbook has been followed by an engineer other than the author with no additional help from the author (verified walkthrough) - [ ] Links to monitoring dashboards and log queries relevant to the procedure are included and verified working ### Recommended - [ ] Each step includes a "what if this fails" note with the next troubleshooting action - [ ] Runbook linked from the alert that triggers it (alert annotation or on-call tool) - [ ] Runbook version history tracked: each update notes what changed and why - [ ] Estimated time for the procedure documented (helps operator manage SLA timers) - [ ] Rollback or undo instructions included for every step that makes a state change ## Verification **Manual steps (all required):** - Independent walkthrough: an engineer who did not write the runbook executes it in a staging or non-production environment and reports any steps that were unclear, missing, or produced unexpected output - Author resolves all feedback from the walkthrough before marking the runbook done - Link verification: at least three runbook links (dashboards, log queries, related runbooks) confirmed reachable ## Tailoring Guide **Add criteria when:** - Deployment runbook: require rollback section meets dod-deployment rollback criteria - Security incident runbook: require evidence collection steps before any remediation steps (preserve logs, capture snapshots) - Runbook for a regulated process: require version-controlled audit trail of each execution (who ran it, when, outcome) - High-complexity procedure: require runbook has a "stop and escalate" decision point after each major phase **Remove or relax criteria when:** - Runbook is a one-time procedure (data backfill, one-off migration): may relax version history requirement; require post-execution notes appended instead - Automated procedure where the runbook is actually a script: document the script as the runbook; require the script tested in staging with expected output captured ## Extension Points - `ext-runbook-template` — organization runbook template with standard sections and formatting - `ext-runbook-review` — runbook review and approval workflow (who must sign off before a runbook is live) - `ext-runbook-testing` — automated runbook test harness (chaos engineering integration)