arn-spark-discover
This skill should be used when the user says "discover", "product discovery", "arn discover", "help me define this product", "what should I build", "product concept", "define the product", "let's figure out what to build", "vision for this project", "shape this idea", "new project idea", "brainstorm this product", "starting from scratch", or wants to explore and structure a greenfield product idea through guided conversation. Produces a product-concept.md document capturing the product vision, core experience, target users, trust model, platforms, and scope boundaries.
59
68%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./plugins/arn-spark/skills/arn-spark-discover/SKILL.mdQuality
Discovery
89%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 is a well-constructed description with excellent trigger term coverage and clear completeness, explicitly stating both what the skill does and when to use it. The main weakness is that the specific capabilities could be more concretely articulated—listing discrete actions rather than describing a general process. The heavy front-loading of trigger phrases, while thorough, makes the description read more like a keyword list than a natural capability statement.
Suggestions
Rephrase to lead with concrete actions (e.g., 'Guides users through product discovery by identifying target users, defining core experience, establishing trust models, and setting scope boundaries') before listing trigger terms.
Consider reducing the trigger phrase list slightly and instead grouping them more naturally (e.g., 'Use when the user wants to define, brainstorm, or shape a new product idea from scratch').
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description names the domain (product discovery) and mentions producing a specific document (product-concept.md) with listed sections (vision, core experience, target users, etc.), but the actual actions are somewhat vague—'explore and structure a greenfield product idea through guided conversation' is more process-oriented than listing concrete discrete actions. | 2 / 3 |
Completeness | The description explicitly answers both 'what' (produces a product-concept.md capturing vision, core experience, target users, trust model, platforms, and scope boundaries through guided conversation) and 'when' (extensive list of trigger phrases and the condition 'wants to explore and structure a greenfield product idea'). | 3 / 3 |
Trigger Term Quality | The description includes an extensive list of natural trigger phrases users would actually say, such as 'what should I build', 'brainstorm this product', 'starting from scratch', 'shape this idea', and 'new project idea'. These cover a wide range of natural language variations. | 3 / 3 |
Distinctiveness Conflict Risk | The skill has a clear niche—greenfield product discovery and concept definition—with very specific trigger terms like 'product discovery', 'arn discover', and 'product concept'. The specific output artifact (product-concept.md) further distinguishes it from general brainstorming or project management skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
47%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill has excellent workflow clarity with well-defined steps, checkpoints, validation loops, and comprehensive error handling. However, it is severely over-long and verbose — the main SKILL.md tries to be both an overview and a detailed reference manual simultaneously, resulting in a massive document that wastes context window. The duplicated situation/action tables and extensive inline detail about conversational strategies and multi-phase orchestration should be extracted into reference files.
Suggestions
Extract the detailed Checkpoint B (persona generation), Checkpoint C (competitive landscape orchestration), and Checkpoint D (assumptions/success criteria) flows into separate reference files, keeping only 2-3 line summaries in the main SKILL.md
Remove the duplicated Agent Invocation Guide table at the bottom — it largely repeats the situation/action table in Step 3 and adds ~30 lines of redundant content
Cut conversational coaching that Claude already knows (e.g., 'Acknowledge with a brief restatement to confirm understanding', 'Do not add interpretation or assumptions beyond what the user stated', 'Brief answers are acceptable if the topic is genuinely simple') to reduce token count by ~20-30%
Show one concrete example of an agent Task tool invocation with actual parameters rather than describing them abstractly — this would improve actionability significantly
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | This skill is extremely verbose at ~350+ lines of detailed instructions. It over-explains conversational strategies, includes extensive decision tables that duplicate each other (the situation/action table appears twice with overlapping content), and provides granular detail about multi-phase orchestration that could be dramatically condensed. Much of the content describes conversational nuance that Claude already understands (how to ask follow-up questions, how to acknowledge answers). | 1 / 3 |
Actionability | The skill provides concrete agent invocation patterns, specific checkpoint triggers, and template references, which is good. However, it lacks executable code/commands — the agent invocations are described abstractly ('invoke via Task tool') without showing exact tool call syntax or parameter structures. The file path references use variables but no concrete examples of the actual calls are shown. | 2 / 3 |
Workflow Clarity | The multi-step workflow is clearly sequenced (Steps 0-5) with explicit validation checkpoints (Checkpoints A-D), a readiness check with status tracking, and error recovery paths for each failure mode. The competitive landscape research has a well-defined three-phase fan-out/fan-in pattern with clear progression gates. The feedback loops (user validates/refines, iterate until satisfied) are explicit throughout. | 3 / 3 |
Progressive Disclosure | The skill references external files appropriately (discovery-questions.md, product-concept-template.md, persona-profile-template.md, competitive-landscape-template.md, step-0-fast-path.md, ensure-config.md), which is good progressive disclosure. However, the SKILL.md itself is a monolithic wall of text that inlines enormous amounts of detail that could be split into reference files — the competitive landscape orchestration, persona generation flow, and checkpoint details could each be separate references, keeping the main skill leaner. | 2 / 3 |
Total | 8 / 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
- AppsVortex/arness
- Commit
b9084b6
- 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.