ralph
Self-referential loop until task completion with configurable verification reviewer
35
31%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./skills/ralph/SKILL.mdQuality
Discovery
0%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This description is extremely vague and uses abstract, technical jargon without conveying any concrete capabilities or use cases. It fails on all dimensions: it doesn't specify what actions the skill performs, includes no natural trigger terms, lacks explicit 'when to use' guidance, and is too generic to be distinguishable from other skills.
Suggestions
Replace abstract language with concrete actions — specify what kind of tasks this skill loops on (e.g., 'Iteratively refines code, documents, or outputs by running repeated review cycles until quality criteria are met').
Add a 'Use when...' clause with natural trigger terms users would say, such as 'Use when the user asks to iterate, refine, review repeatedly, or keep improving until done'.
Clarify what 'configurable verification reviewer' means in practical terms — e.g., 'Supports custom review criteria or checklist-based verification at each iteration step'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description uses vague, abstract language like 'self-referential loop' and 'task completion' without specifying any concrete actions. It does not describe what the skill actually does in practical terms. | 1 / 3 |
Completeness | The description fails to clearly answer 'what does this do' in concrete terms and completely lacks any 'when should Claude use it' guidance. There is no 'Use when...' clause or equivalent. | 1 / 3 |
Trigger Term Quality | The terms 'self-referential loop', 'configurable verification reviewer' are technical jargon that no user would naturally say. There are no natural keywords a user would use to invoke this skill. | 1 / 3 |
Distinctiveness Conflict Risk | The description is so vague that it's unclear what domain or task type it applies to. 'Task completion' could apply to virtually any skill, creating high conflict risk with other skills. | 1 / 3 |
Total | 4 / 12 Passed |
Implementation
62%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a comprehensive, highly actionable workflow skill with excellent step sequencing, validation checkpoints, and concrete examples. Its main weakness is verbosity — it over-explains rationale, anti-patterns, and context that Claude doesn't need, roughly doubling the token cost. The monolithic structure could benefit from splitting detailed sections (examples, tool usage, advanced rules) into referenced files.
Suggestions
Cut the <Why_This_Exists>, <Use_When>, and <Do_Not_Use_When> sections significantly — these are selection hints that belong in frontmatter or a brief one-liner, not multi-line explanations.
Move the <Examples> and <Advanced> sections into separate referenced files (e.g., EXAMPLES.md, ADVANCED.md) to reduce the main skill's token footprint.
Remove repeated warnings about the 'polite-stop anti-pattern' — stating it once in the Steps section is sufficient; it appears three times across Steps, Escalation, and Tool_Usage.
Trim explanatory 'Why good/Why bad' annotations in examples to single-line comments — Claude can infer the reasoning from the contrast.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~200+ lines. Contains extensive explanations of why Ralph exists, when to use/not use it, detailed rationale sections, and repeated warnings about anti-patterns. Much of this (e.g., explaining what PRD-driven development is, why partial implementations fail, what 'polite-stop anti-pattern' means) is context Claude doesn't need spelled out at this length. The same content could be conveyed in roughly half the tokens. | 1 / 3 |
Actionability | Highly actionable with concrete steps, specific tool invocation syntax (Task, Skill, omc ask codex), exact file paths (.omc/state/sessions/{sessionId}/prd.json), specific commands, and detailed examples showing both correct and incorrect patterns. The PRD refinement example with before/after JSON is particularly executable. | 3 / 3 |
Workflow Clarity | Excellent multi-step workflow with clear sequencing (Steps 1-9 including 7.5 and 7.6), explicit validation checkpoints at multiple stages (story verification, reviewer verification, regression re-verification), feedback loops for error recovery (rejection → fix → re-verify), and a final checklist. The explicit anti-pattern warnings about stopping prematurely after Step 7 add clarity to the flow. | 3 / 3 |
Progressive Disclosure | References external files (docs/shared/agent-tiers.md, docs/company-context-interface.md, .claude/omc.jsonc) but the SKILL.md itself is monolithic — all content is inline rather than split into overview + detailed references. The Examples, Advanced, and Tool_Usage sections could be separate files. No bundle files are provided to verify referenced paths exist. | 2 / 3 |
Total | 9 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 10 / 11 Passed | |
- Repository
- Yeachan-Heo/oh-my-claudecode
- Commit
679b418
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.