# Rune Refiner > Analyze RUNE specifications and suggest improvements. Use when a spec is valid but may have missing tests, uncovered edge cases, ambiguous behavior rules, or incomplete constraints. Use when this capability is needed. - Skill: `tomevault-io/rune-refiner` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/rune-refiner` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/rune-refiner/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/rune-refiner --- ## How to Use When given a RUNE spec that is structurally valid but may be incomplete or weak, analyze it and suggest improvements with ready-to-use content. Input: a `.rune` YAML file or Markdown spec section. Output: a list of actionable suggestions with proposed content. --- ## Analysis Areas ### 1. Test Coverage Gaps Check if the TESTS section adequately covers the BEHAVIOR rules. **Look for:** - BEHAVIOR rules with only 1 test (should have 2+) - Missing boundary tests for numeric ranges - Missing empty/null input tests - Missing tests for the OTHERWISE clause - Happy path tests that only cover one scenario **Example suggestion:** ```markdown **Gap:** BEHAVIOR rule "WHEN percentage > 100 THEN raise error" has only 1 test with value 101. **Suggestion:** Add a test with a more extreme value: - `"calculate_discount(100.0, 150) raises ValueError"` ``` ### 2. Edge Case Discovery Identify edge cases the spec author may have missed. **Common edge cases by input type** (adapt null/empty representations to the target language): | Type | Edge cases to check | |------|-------------------| | string | empty `""`, whitespace-only, very long (>1000 chars), unicode, null/nil/None | | integer | 0, negative, max value for the language, null/nil/None | | decimal/float | 0.0, negative, very small (0.001), very large, NaN, Infinity, null/nil/None | | list/array | empty, single item, duplicate items, null/nil/None | | map/object | empty, missing keys, extra keys, null/nil/None | | boolean | both true and false | | date/time | today, past, far future, midnight, timezone boundaries | **Example suggestion:** ```markdown **Gap:** No edge case for whitespace-only input. **Suggestion:** Add to EDGE_CASES: - `"whitespace-only email (' '): returns (False, 'Email cannot be empty')"` And add to TESTS: - `"validate_email(' ')[0] == False"` ``` ### 3. BEHAVIOR Ambiguity Identify behavior rules that are vague or open to multiple interpretations. **Red flags:** - Rules without specific error messages: `THEN raise error` - Rules with vague conditions: `WHEN input is invalid` - Rules that use "etc." or "and so on" - Missing OTHERWISE clause - Conditions that could overlap (multiple rules match the same input) **Example suggestion:** ```markdown **Ambiguity:** "WHEN input is invalid THEN raise error" — what makes input invalid? What error? **Suggestion:** Replace with specific rules: - `WHEN input is empty THEN raise error "Input cannot be empty"` - `WHEN input contains non-ASCII characters THEN raise error "Input must be ASCII"` ``` ### 4. CONSTRAINTS Completeness Check if CONSTRAINTS fully describe the valid input space. **Look for:** - Parameters without any constraint - Constraints that are vague ("must be valid") - Missing range information for numeric parameters - Missing format information for string parameters - Missing size limits for collections **Example suggestion:** ```markdown **Gap:** Parameter `timeout` has no constraint. **Suggestion:** Add to CONSTRAINTS: - `"timeout: integer between 1 and 300 (seconds)"` ``` ### 5. INTENT Clarity Check if INTENT clearly communicates purpose. **Issues to flag:** - Implementation details (mentioning regex, algorithms, libraries) - Too vague ("processes data") - Missing return value description - More than 3 sentences ### 6. Spec Consistency Check internal consistency across sections. **Look for:** - EDGE_CASES that aren't reflected in BEHAVIOR - CONSTRAINTS that imply behavior not in BEHAVIOR - EXAMPLES that show behavior not in BEHAVIOR - COMPLEXITY that doesn't match the apparent algorithm --- ## Output Format ```markdown ## RUNE Refinement Report: `` ### Test Coverage 1. **[ADD TEST]** BEHAVIOR rule "WHEN percentage > 100..." has 1 test. Add: - `"calculate_discount(100.0, 200) raises ValueError"` 2. **[ADD TEST]** OTHERWISE clause has no dedicated test. Add: - `"calculate_discount(100.0, 20) == 80.0"` (if not already present) ### Edge Cases 3. **[ADD EDGE_CASE]** No test for zero price: - Edge case: `"price = 0.0: returns 0.0 regardless of percentage"` - Test: `"calculate_discount(0.0, 50) == 0.0"` 4. **[ADD EDGE_CASE]** No test for float precision: - Edge case: `"rounding: 10.0 * 33% = 6.70, not 6.7"` - Test: `"calculate_discount(10.0, 33) == 6.70"` ### Ambiguity 5. **[CLARIFY]** BEHAVIOR rule "WHEN coupon has expired" — expired means `expires_at < current_date` or `expires_at <= current_date`? Add to EDGE_CASES: - `"expires today: still valid (expires_at == current_date)"` ### Constraints 6. **[ADD CONSTRAINT]** Parameter `date` has no format constraint. Add: - `"date: ISO 8601 string (YYYY-MM-DD)"` ### Summary - **Suggestions:** 6 total (2 tests, 2 edge cases, 1 clarification, 1 constraint) - **Estimated strength improvement:** The spec goes from covering ~70% of realistic inputs to ~95% ``` --- ## Rules - Only suggest improvements, never modify the spec directly - Every suggestion must include ready-to-paste content (YAML or Markdown) - Prioritize suggestions that prevent real bugs over stylistic improvements - Do not suggest changes that contradict existing BEHAVIOR rules - If the spec is already strong, say so — don't invent unnecessary suggestions - Focus on gaps that would cause different AI tools to generate different behavior --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/vict00r99) — claim your Tome and manage your conversions.