From arn-spark
This skill should be used when the user says "prfaq", "pr faq", "pr/faq", "press release stress test", "stress prfaq", "amazon pr faq method", "test the pitch with a pr/faq", "validate concept through pr/faq", "critique press release", "pr faq stress test", "will this marketing story hold up", or wants to stress-test a product concept by drafting a compelling press release and FAQ, then adversarially critiquing it to find where the concept cracks under scrutiny. Produces a PR/FAQ report with the full draft, adversarial questions, crack point analysis, and recommended concept updates.
npx claudepluginhub appsvortex/arness --plugin arn-sparkThis skill uses the workspace's default tool permissions.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Implements structured self-debugging workflow for AI agent failures: capture errors, diagnose patterns like loops or context overflow, apply contained recoveries, and generate introspection reports.
Stress-test a product concept using Amazon's PR/FAQ method. This technique forces the product concept through two filters:
The two phases use separate agent invocations to prevent rubber-stamping. A critic who remembers being the drafter unconsciously defends what it wrote. Separate invocations force genuine adversarial evaluation.
The product concept is read but never modified -- all recommendations are captured in the PR/FAQ report for later review.
CLAUDE.md and check for a ## Arness section## Arness section exists or Arness Spark fields are missing, inform the user: "Arness Spark is not configured for this project yet. Run /arn-brainstorming to get started — it will set everything up automatically." Do not proceed without it.mkdir -p <reports-dir>/stress-tests/| Artifact | Status | Location | Fallback |
|---|---|---|---|
| Product concept | REQUIRED | <vision-dir>/product-concept.md | Cannot proceed without it -- suggest running /arn-spark-discover |
| Product pillars | ENRICHES | Product Pillars section of product concept | Draft messaging is less focused; critique has fewer anchors |
| Competitive landscape | ENRICHES | Competitive Landscape section of product concept | Draft positioning is less grounded in market context |
| Target personas | ENRICHES | Target Personas section of product concept | Customer quote in press release is less persona-specific |
Standard fallback cascade:
If no product concept exists:
Ask (using AskUserQuestion): "No product concept found. The PR/FAQ stress test needs a product concept to draft and critique messaging for. How would you like to proceed?"
/arn-spark-discover to create a product concept firstIf the user chooses option 2, collect a product description and proceed with a reduced-fidelity test (note in the report that the test was based on a verbal description rather than a full product concept).
Load the PR/FAQ workflow and report template:
Read
${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-stress-prfaq/references/prfaq-workflow.mdRead${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-stress-prfaq/references/prfaq-report-template.md
Read the product concept from <vision-dir>/product-concept.md. Extract:
Invoke arn-spark-marketing-pm in draft mode with:
--- PRODUCT CONCEPT --- [full product concept document] --- END PRODUCT CONCEPT ---
--- PRODUCT PILLARS --- [product pillars section] --- END PRODUCT PILLARS ---
--- OPERATING MODE --- draft --- END OPERATING MODE ---
Receive back the complete draft: press release (400-600 words), customer FAQ (5-8 entries), internal FAQ (3-5 entries).
Quality check before proceeding:
If the draft is too thin or generic, retry with more specific context:
"The draft needs to be more specific. Ground it in these details:
Invoke arn-spark-marketing-pm in critique mode with:
--- PRODUCT CONCEPT --- [full product concept document -- same as Phase 1] --- END PRODUCT CONCEPT ---
--- PRODUCT PILLARS --- [product pillars section -- same as Phase 1] --- END PRODUCT PILLARS ---
--- OPERATING MODE --- critique --- END OPERATING MODE ---
--- DRAFT OUTPUT --- [complete draft from Phase 1 -- press release + customer FAQ + internal FAQ] --- END DRAFT OUTPUT ---
Critical: This MUST be a separate agent invocation from Phase 1. Include ONLY: the full product concept, the product pillars, the complete draft output (press release + FAQs), and the critique task instructions. Do NOT include any conversation context from the draft phase or any conversational recap.
Receive back: adversarial questions (5-8), crack points (3-5), recommended concept updates table, unresolved questions.
Quality check:
If the critique is too soft, retry with explicit adversarial instruction:
"The critique needs to be sharper. Requirements:
Review the critique's Recommended Concept Updates table. Ensure:
If the critique's recommendations are incomplete, supplement from the crack point "What needs strengthening" fields.
Using the PR/FAQ report template:
<reports-dir>/stress-tests/prfaq-report.mdPresent a summary to the user:
"PR/FAQ stress test complete. Report saved to [path].
Draft assessment: [1 sentence on whether the product story was compelling]
Key critique findings:
Top crack point: [The highest-impact crack point in 1 sentence]
Recommended concept updates: [N] recommendations ([X] Add, [Y] Modify, [Z] Remove) Unresolved questions: [N]
This report will be used by /arn-spark-concept-review to propose changes to the product concept."
| Situation | Agent | Mode/Context |
|---|---|---|
| Write press release and FAQ | arn-spark-marketing-pm | Draft mode with product concept and pillars |
| Adversarially critique the draft | arn-spark-marketing-pm | Critique mode with product concept, pillars, and draft output (separate invocation) |
Marketing PM produces generic/thin draft: Retry with more specific context from the product concept -- highlight specific features, personas, and competitive positioning. If retry still produces thin output:
Ask (using AskUserQuestion): "The PR/FAQ draft is too generic to produce a useful critique. How would you like to proceed?"
Marketing PM produces soft critique: Retry with explicit adversarial instruction emphasizing that the critique must target concept substance, not copywriting quality. Include: "If no crack point makes someone uncomfortable, the critique has failed."
Critique mode receives draft context (accidental context leak): This should not happen if invocations are properly separated. If detected (the critique references drafting decisions or uses phrases like "when I wrote..."), discard the critique and re-invoke in critique mode with only the draft output and product concept -- no conversational context.
Any agent invocation fails entirely: Retry once with a simplified prompt. If retry fails:
Ask (using AskUserQuestion): "Agent invocation failed. How would you like to proceed?"
prfaq-report.md already exists, it is overwritten. Git provides history.