ai-toolkit-rules

Name: ai-toolkit-rules
Author: softspark

161stars21forks

How to install ai-toolkit-rules?

softspark/ai-toolkit/ai-toolkit-rules

$npx -y skills add softspark/ai-toolkit --skill ai-toolkit-rules

Installs into the current project.

Summary

Mandatory engineering, security, testing, git, performance, quality, and response rules. Claude MUST load this skill for every technical, coding, debugging, review, architecture, DevOps, data, or file-editing task in Chat or Cowork.

Files · 1

View on GitHub

SKILL.md

1	# AI Toolkit Rules
2
3	Apply every relevant rule below before acting. Treat MUST/NEVER language as mandatory.
4
5	## Source: `app/rules/claude-toolkit-rules.md`
6
7	# Claude Toolkit
8
9	Shared AI development toolkit — lifecycle hooks, safety constitution, multi-platform support.
10
11	## Skill Tiers
12
13	- Tier 1 — single-agent: `/debug`, `/review`, `/refactor`, `/analyze`, `/docs`, `/plan`, `/explain`, `/tdd`, `/triage-issue`
14	- Tier 1.5 — planning: `/write-a-prd` → `/prd-to-plan` → `/prd-to-issues`; design: `/design-an-interface`, `/architecture-audit`, `/refactor-plan`
15	- Tier 2 — multi-agent: `/workflow <type>` (feature-development, backend-feature, frontend-feature, api-design, database-evolution, test-coverage, security-audit, debugging, incident-response, spike, codebase-onboarding, performance-optimization, infrastructure-change, application-deploy, proactive-troubleshooting)
16	- Tier 3 — custom: `/orchestrate <desc>` (3–6 agents) \| `/swarm <mode> <desc>` (map-reduce \| consensus \| relay)
17
18	## Path Safety
19	- NEVER guess or hallucinate user home directory paths
20	- Use `~` or `$HOME` instead of hardcoded `/Users/<username>/` or `/home/<username>/`
21	- When an absolute path is needed, run `echo $HOME` first to get the correct value
22
23	## User Preferences
24
25	- Style: Direct & efficient. No pleasantries. Measurable results.
26	- Methodology: Provide >=3 alternatives. Use Socratic questioning.
27	- Review: Apply "Devil's Advocate" critique to decisions.
28
29	## Source: `app/rules/git-conventions.md`
30
31	# Git Conventions
32
33	- Do NOT add `Co-Authored-By: Claude` or any AI co-authorship to commits
34	- Do NOT add Claude signatures or attribution to commit messages
35	- Conventional commits format: `feat:`, `fix:`, `docs:`, `refactor:`, `test:`, `chore:`
36
37	## Source: `app/rules/output-mode.md`
38
39	# Output Mode
40
41	`output-mode: concise`
42
43	Default response mode for this project is concise. The `brand-voice` skill (when present in ai-toolkit) auto-loads its `concise` rules; assistants without that skill should still apply the directives below.
44
45	## Concise Mode Directives
46
47	- No preamble. Skip "I'll now...", "Sure, let me...", "Great question!" and similar warm-ups. Start with the answer.
48	- Lead with the result. Conclusion or output first; explanation only if asked or non-obvious.
49	- Max 3 sentences per closed question. Yes/no, single-fact, or "where is X" answers stay under three sentences.
50	- Tables and lists over prose when comparing options, listing steps, or showing values.
51	- No trailing summaries. If the diff or output already shows what changed, do not restate it.
52	- Drop filler adjectives. No "nice", "great", "powerful", "robust" unless the user asked for evaluation.
53	- Cite file paths as `path:line` instead of paragraphs describing where things live.
54	- Reserve longer prose for: architecture proposals, trade-off analyses, plans with risks. Everything else: terse.
55
56	## When to escalate to verbose
57
58	- User explicitly asks: "explain in detail", "walk me through", "give me the full picture".
59	- Reporting a non-obvious failure mode where missing context would mislead.
60	- Architecture / RFC / ADR / trade-off documents — those have their own structure.
61
62	## How to override
63
64	- Per-session: `/brand-voice default` (or `/brand-voice strict` for even tighter)
65	- Per-project: change this rule's `output-mode:` value in the project's `CLAUDE.md`
66	- Permanent removal: re-run `ai-toolkit install --skip rules` or strip the `<!-- TOOLKIT:output-mode -->` block manually
67
68	## Source: `app/rules/quality-gates.md`
69
70	# Quality Gates & Mandatory Practices
71
72	## MANDATORY PRACTICES
73	1. Plan First: Tasks >1h require Plan, Success Criteria, and Pre-Mortem.
74	2. Quality Gates:
75	* `ruff check .` (0 errors)
76	* `mypy --strict src/` (0 errors)
77	* `pytest --cov=src` (>70% coverage)
78	* Type Safety: 100% public APIs, >60% internal.
79	3. Security: No secrets in code, sanitization, auth z/n.
80
81	## Source: `app/rules/common/coding-style.md`
82
83	# Universal Coding Style
84
85	## Principles
86	- KISS: simplest solution that works. Clever code is a liability. If 200 lines could be 50, rewrite.
87	- DRY: extract when you repeat 3+ times, not before.
88	- YAGNI: do not build features "just in case." No abstractions for single-use code.
89	- Prefer immutability: use `const`, `final`, `val`, `let` by default.
90	- Fail fast: validate inputs at boundaries, return early on errors.
91	- State assumptions before coding. If uncertain or multiple interpretations exist, ask — don't pick silently.
92
93	## Naming
94	- Use descriptive names that reveal intent (`remainingRetries`, not `r`).
95	- Boolean variables/functions: prefix with `is`, `has`, `can`, `should`.
96	- Functions: verb

People who liked this also viewed

6 picks

Type

ai-toolkit-rules

161stars21forks

View on GitHub

How to install ai-toolkit-rules?

softspark/ai-toolkit/ai-toolkit-rules

$npx -y skills add softspark/ai-toolkit --skill ai-toolkit-rules

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# AI Toolkit Rules
2
3	Apply every relevant rule below before acting. Treat MUST/NEVER language as mandatory.
4
5	## Source: `app/rules/claude-toolkit-rules.md`
6
7	# Claude Toolkit
8
9	Shared AI development toolkit — lifecycle hooks, safety constitution, multi-platform support.
10
11	## Skill Tiers
12
13	- Tier 1 — single-agent: `/debug`, `/review`, `/refactor`, `/analyze`, `/docs`, `/plan`, `/explain`, `/tdd`, `/triage-issue`
14	- Tier 1.5 — planning: `/write-a-prd` → `/prd-to-plan` → `/prd-to-issues`; design: `/design-an-interface`, `/architecture-audit`, `/refactor-plan`
15	- Tier 2 — multi-agent: `/workflow <type>` (feature-development, backend-feature, frontend-feature, api-design, database-evolution, test-coverage, security-audit, debugging, incident-response, spike, codebase-onboarding, performance-optimization, infrastructure-change, application-deploy, proactive-troubleshooting)
16	- Tier 3 — custom: `/orchestrate <desc>` (3–6 agents) \| `/swarm <mode> <desc>` (map-reduce \| consensus \| relay)
17
18	## Path Safety
19	- NEVER guess or hallucinate user home directory paths
20	- Use `~` or `$HOME` instead of hardcoded `/Users/<username>/` or `/home/<username>/`
21	- When an absolute path is needed, run `echo $HOME` first to get the correct value
22
23	## User Preferences
24
25	- Style: Direct & efficient. No pleasantries. Measurable results.
26	- Methodology: Provide >=3 alternatives. Use Socratic questioning.
27	- Review: Apply "Devil's Advocate" critique to decisions.
28
29	## Source: `app/rules/git-conventions.md`
30
31	# Git Conventions
32
33	- Do NOT add `Co-Authored-By: Claude` or any AI co-authorship to commits
34	- Do NOT add Claude signatures or attribution to commit messages
35	- Conventional commits format: `feat:`, `fix:`, `docs:`, `refactor:`, `test:`, `chore:`
36
37	## Source: `app/rules/output-mode.md`
38
39	# Output Mode
40
41	`output-mode: concise`
42
43	Default response mode for this project is concise. The `brand-voice` skill (when present in ai-toolkit) auto-loads its `concise` rules; assistants without that skill should still apply the directives below.
44
45	## Concise Mode Directives
46
47	- No preamble. Skip "I'll now...", "Sure, let me...", "Great question!" and similar warm-ups. Start with the answer.
48	- Lead with the result. Conclusion or output first; explanation only if asked or non-obvious.
49	- Max 3 sentences per closed question. Yes/no, single-fact, or "where is X" answers stay under three sentences.
50	- Tables and lists over prose when comparing options, listing steps, or showing values.
51	- No trailing summaries. If the diff or output already shows what changed, do not restate it.
52	- Drop filler adjectives. No "nice", "great", "powerful", "robust" unless the user asked for evaluation.
53	- Cite file paths as `path:line` instead of paragraphs describing where things live.
54	- Reserve longer prose for: architecture proposals, trade-off analyses, plans with risks. Everything else: terse.
55
56	## When to escalate to verbose
57
58	- User explicitly asks: "explain in detail", "walk me through", "give me the full picture".
59	- Reporting a non-obvious failure mode where missing context would mislead.
60	- Architecture / RFC / ADR / trade-off documents — those have their own structure.
61
62	## How to override
63
64	- Per-session: `/brand-voice default` (or `/brand-voice strict` for even tighter)
65	- Per-project: change this rule's `output-mode:` value in the project's `CLAUDE.md`
66	- Permanent removal: re-run `ai-toolkit install --skip rules` or strip the `<!-- TOOLKIT:output-mode -->` block manually
67
68	## Source: `app/rules/quality-gates.md`
69
70	# Quality Gates & Mandatory Practices
71
72	## MANDATORY PRACTICES
73	1. Plan First: Tasks >1h require Plan, Success Criteria, and Pre-Mortem.
74	2. Quality Gates:
75	* `ruff check .` (0 errors)
76	* `mypy --strict src/` (0 errors)
77	* `pytest --cov=src` (>70% coverage)
78	* Type Safety: 100% public APIs, >60% internal.
79	3. Security: No secrets in code, sanitization, auth z/n.
80
81	## Source: `app/rules/common/coding-style.md`
82
83	# Universal Coding Style
84
85	## Principles
86	- KISS: simplest solution that works. Clever code is a liability. If 200 lines could be 50, rewrite.
87	- DRY: extract when you repeat 3+ times, not before.
88	- YAGNI: do not build features "just in case." No abstractions for single-use code.
89	- Prefer immutability: use `const`, `final`, `val`, `let` by default.
90	- Fail fast: validate inputs at boundaries, return early on errors.
91	- State assumptions before coding. If uncertain or multiple interpretations exist, ask — don't pick silently.
92
93	## Naming
94	- Use descriptive names that reveal intent (`remainingRetries`, not `r`).
95	- Boolean variables/functions: prefix with `is`, `has`, `can`, `should`.
96	- Functions: verb

People who liked this also viewed

6 picks

Type