Skip to content

Commit 76e71ed

Browse files
authored
Add k8s tooling and extend Claude sample rules (#9)
* Add Kubernetes tooling: helm, kubectx, kluctl, kind, ko, oras * Extend Claude sample rules: thinking, simplicity, surgical edits, goals, tooling Add behavioral guidance covering pre-implementation reasoning, minimal implementations, scoped edits, verifiable success criteria, and tool preferences (yq for YAML, draft PRs by default).
1 parent 9a5028e commit 76e71ed

4 files changed

Lines changed: 72 additions & 2 deletions

File tree

AGENTS.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -55,7 +55,7 @@ The repository provides a two-step setup process:
5555
- Detects platform (macOS only currently)
5656
- Prompts `[Y/n]` for each tool (Enter = yes, `n` = skip)
5757
- Installs Homebrew if missing
58-
- Installs packages: stow, k9s, kubectl, awscli, gh, jq, yq, ncdu, git, go, mise, gitleaks
58+
- Installs packages: stow, k9s, kubectl, kubectx, helm, kind, ko, oras, kluctl, awscli, gh, jq, yq, ncdu, git, go, mise, gitleaks
5959
- Installs casks: ghostty (terminal emulator), raycast (productivity launcher), cursor (AI code editor), docker (Docker Desktop)
6060
- Installs Rust via rustup
6161
- Installs oh-my-zsh with plugins (zsh-autosuggestions, zsh-syntax-highlighting)

Brewfile

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -7,6 +7,12 @@ brew "ncdu"
77
brew "go"
88
brew "k9s"
99
brew "kubectl"
10+
brew "kubectx"
11+
brew "helm"
12+
brew "kind"
13+
brew "ko"
14+
brew "oras"
15+
brew "kluctl/tap/kluctl"
1016
brew "awscli"
1117
brew "mise"
1218
brew "gitleaks"

configs/claude/.claude/CLAUDE.sample.md

Lines changed: 64 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -40,3 +40,67 @@ Always find the root cause. Never fix a symptom or add a workaround.
4040
# Comments
4141

4242
When writing code comments, describe "why" not "what". Never make descriptive comments that redundantly encode what can trivially be understood by reading well-named variables and functions.
43+
44+
# Think Before Coding
45+
46+
Don't assume. Don't hide confusion. Surface tradeoffs.
47+
48+
Before implementing:
49+
50+
- State your assumptions explicitly. If uncertain, ask.
51+
- If multiple interpretations exist, present them — don't pick silently.
52+
- If a simpler approach exists, say so. Push back when warranted.
53+
- If something is unclear, stop. Name what's confusing. Ask.
54+
55+
# Simplicity First
56+
57+
Minimum code that solves the problem. Nothing speculative.
58+
59+
- No features beyond what was asked.
60+
- No abstractions for single-use code.
61+
- No "flexibility" or "configurability" that wasn't requested.
62+
- No error handling for impossible scenarios.
63+
- If you write 200 lines and it could be 50, rewrite it.
64+
65+
Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.
66+
67+
# Surgical Changes
68+
69+
Touch only what you must. Clean up only your own mess.
70+
71+
When editing existing code:
72+
73+
- Don't "improve" adjacent code, comments, or formatting.
74+
- Don't refactor things that aren't broken.
75+
- Match existing style, even if you'd do it differently.
76+
- If you notice unrelated dead code, mention it — don't delete it.
77+
78+
When your changes create orphans:
79+
80+
- Remove imports/variables/functions that YOUR changes made unused.
81+
- Don't remove pre-existing dead code unless asked.
82+
83+
The test: every changed line should trace directly to the user's request.
84+
85+
# Goal-Driven Execution
86+
87+
Define success criteria. Loop until verified.
88+
89+
Transform tasks into verifiable goals:
90+
91+
- "Add validation" → "Write tests for invalid inputs, then make them pass"
92+
- "Fix the bug" → "Write a test that reproduces it, then make it pass"
93+
- "Refactor X" → "Ensure tests pass before and after"
94+
95+
For multi-step tasks, state a brief plan:
96+
97+
1. [Step] → verify: [check]
98+
2. [Step] → verify: [check]
99+
3. [Step] → verify: [check]
100+
101+
Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.
102+
103+
# Tools Usage
104+
105+
- For YAML file validation, always use `yq`. Never write a Python script for verification.
106+
- When opening a GitHub pull request, always open it as a draft unless directly asked to open a normal pull request.

install.sh

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,7 @@ echo "✅ Platform: macOS"
1919

2020
# ── Item registry ──────────────────────────────────────────────────
2121

22-
add_item "homebrew-tools" "16 packages via Brewfile"
22+
add_item "homebrew-tools" "22 packages via Brewfile"
2323
add_item "rust" "via rustup"
2424
add_item "oh-my-zsh" "zsh framework"
2525
add_item "nvm" "Node.js version manager"

0 commit comments

Comments
 (0)