← ClaudeAtlas

devopslisted

DevOps toolkit for a small platform team, generalised from a Thai-government integration project: (1) change — any infra work that touches a real machine (ssh via a jump host, kubectl, HAProxy / ingress / vault-backed secrets, incident root cause) done in a fixed order and closed with a before / steps / after / rollback report with a SEV level; (2) release-scope — what goes up in the next release and what is still pending, from git keys across the product's repos plus the issue boards, written as the PM list and checked against the customer message; (3) network-diagram — the team's system / network diagram as one draw.io file with a board per version, exported to a clean one-page PDF, rebuilt from a PDF, benchmarked. Use WHENEVER the work is "ssh to the server", "edit haproxy", "root cause / postmortem", "what's in release X", "release note for the PM", "update the system diagram", "export the drawio", "เขียน rootcause", "สรุป release ให้ PM", "แก้ diagram ระบบ". Read the sub-skill's SKILL.md before acting.
th3r-dev/th3-skills · ★ 2 · DevOps & Infrastructure · score 66
Install: claude install-skill th3r-dev/th3-skills
# devops — sub-skills | When the work is… | Read | Scripts | |---|---|---| | "ssh to <ip>", "kubectl on the cluster", "edit haproxy.cfg", "root cause", "before/after/rollback report" | [`change/SKILL.md`](change/SKILL.md) + [`references/`](change/references/) (change template TH/EN, incident template) | `change/scripts/jssh.py` (jump-host ssh from an ini, `--sudo` `--kube` `--forward`) · `snapshot_haproxy.sh` | | "what's in v1.2.x", "pending for the next release", "summary for the PM", "check the PM's customer message" | [`release-scope/SKILL.md`](release-scope/SKILL.md) | `release-scope/scripts/release_scope.py` (`--repo` × N, `--key-regex`, `--board` × N → `scope.md` + `scope.json`) | | "update the system diagram", "export the drawio to one PDF", "rebuild the drawio from the PDF", "compare with the final PDF" | [`network-diagram/SKILL.md`](network-diagram/SKILL.md) | `network-diagram/scripts/drawio_board.py` (`list` `clone` `export` `trace` `compare`) + `assets/icons/` | They chain: a `change/` report links the diagram that explains it (`rootcause.html`, or a version board from `network-diagram/`); `release-scope/` tells `change/` which tag a prod deploy carries. ## Rules every sub-skill keeps - **Read the project's own `docs/`, `changes/` and `incidents/` before guessing about a machine.** Facts about your fleet beat any generic cloud / Kubernetes advice. - **Passwords never appear on a command line, in a report, or in chat.** `jssh.py` reads them from `~/.config/d