instruction generator

Other Community

Skill notes

Quick Decision: Which instruction file?

Need · File · Location Applies to the whole workspace / all files · copilot-instructions.md or AGENTS.md · .github/ or root Applies to specific file types or folders · <name>.instructions.md · .github/instructions/ Applies to user (cross-workspace) · <name>.instructions.md · {{VSCODE_USER_PROMPTS_FOLDER}}/ applyTo patterns:

Workflow

Follow these steps in order. Mark each ✓ when done.

Step 1 — Clarify scope

Answer before writing: Who is this for? All team members (workspace) or just this user (user-level)? What files trigger it? Language, framework, folder, or all files? What behavior changes? Style, safety rules, prohibited patterns, or workflow steps?

Step 2 — Choose depth

Variant · Use when · Target length Concise rules · Simple preferences (naming, style, imports) · ≤ 20 lines Extended rules · Fragile workflows, multi-step processes, error-prone areas · 20–60 lines Reference-based · Large rule sets; use progressive disclosure · SKILL body → details.md If the instruction body would exceed 60 lines, split it:

Step 3 — Write the instruction

See templates/instructions-concise.md and templates/instructions-extended.md. To scaffold a stub from the command line:

Fluxly profile Runtime and permissions are inferred by Fluxly static checks, not author declarations or runtime test results.
Domain
Other
Author-declared agents
No explicit declaration found; this is not inferred or tested compatibility
Static check
88 / 100 · heuristic scan, not runtime safety proof
Runtime evidence
Not run; no isolated runtime test evidence is currently recorded
Author / version / license
@tomevault-io · no license declared
Token usage
Lean
Setup complexity
Plug-and-play
External API key
Not required
Operating systems
Unspecified (assume cross-platform)
Runtime requirements
Python
Detected file/system behavior
  • Read-only
  • Write / modify
Network behavior
Local-only
Install commands
None (reference only)

Heads up: Tool permissions are not declared; actual access depends on the runtime.

01
Step 1 — Clarify scope

Answer before writing: Who is this for? All team members (workspace) or just this user (user-level)? What files trigger it? Language, framework, folder, or all files? What behavior changes? Style, safety rules, prohibited patterns, or workflow steps?

02
Step 2 — Choose depth

Variant · Use when · Target length Concise rules · Simple preferences (naming, style, imports) · ≤ 20 lines Extended rules · Fragile workflows, multi-step processes, error-prone areas · 20–60 lines Reference-based · Large rule sets; use progressive disclosure · SKILL body → `details.md` If the instruction body would exceed 60 lines, split it:

03
Step 3 — Write the instruction

See templates/instructions-concise.md and templates/instructions-extended.md. To scaffold a stub from the command line:

04
Step 4 — Rewrite vague rules

Run each rule through this test: "Can Claude determine pass/fail without judgment?" Vague (reject) · Specific (accept) "Write good code" · "Prefer pure functions; avoid side effects in utility modules" "Be concise" · "Keep function bodies under 40 lines; extract helpers if exceeded" "Handle errors properly" · "Wrap async calls in try/catch; surface errors to the caller — never swallow"

05
Step 5 — Check for contradictions

[ ] Does any rule contradict another in this file? [ ] Does any rule contradict existing instructions in `.github/copilot-instructions.md`? [ ] Are `applyTo` patterns precise enough? (Overly broad patterns → high context cost)

06
Step 6 — Validate quality checklist

[ ] Every rule is imperative and verifiable (no "should", "try to", "consider"). [ ] `applyTo` pattern is correct glob syntax — validated against target file paths. [ ] No time-sensitive information (no version numbers with expiry dates, no API phases). [ ] Consistent terminology throughout (same term for same concept). [ ] Instructions don't duplicate VS Code / Copilot defaults.

Discussion

Powered by GitHub Discussions. Sign in with GitHub to comment, react, or subscribe.