Prfaq Beagle

Use when the user wants to pressure-test a product, internal-tool, or OSS concept against Amazon's Working Backwards PRFAQ gauntlet before committing to a sp...

This skill is internally consistent with its stated purpose, but before enabling it you should: 1) be aware it will scan your repository (auto-discover docs) and create/modify files under .beagle/concepts/<slug>/ — back up any sensitive content you don't want scanned or committed; 2) expect it to reuse prior analysis/research outputs by default (it won't overwrite without explicit refresh) — review .beagle before re-running; 3) note that web grounding is performed by companion web-research (which may make web requests or require web tools/credentials) — review the companion's permissions/policy if you care about outbound network access; 4) this skill explicitly forbids writing code or scaffolding, so it won't start builds or deployments. If any of those behaviors are unacceptable in your environment (repo scanning, local file writes, or companion web access), run it in a sanitized or disposable workspace or inspect the companion skills first.

Type: OpenClaw Skill Name: prfaq-beagle Version: 1.0.0 The prfaq-beagle skill bundle implements a structured 'Working Backwards' product discovery process based on Amazon's PRFAQ methodology. It guides an AI agent through five stages (Ignition, Press Release, Customer FAQ, Internal FAQ, and Verdict) to rigorously filter product concepts. The skill utilizes companion skills (artifact-analysis and web-research) to ground its feedback in local documents and market data, following a well-defined 'Companion Invocation Contract' in companion-contract.md. The instructions in SKILL.md and the reference files emphasize a skeptical, 'hardcore' coaching tone but include explicit safety boundaries, such as prohibiting the agent from writing code or scaffolding projects. No indicators of data exfiltration, malicious execution, or unauthorized persistence were found.