Review Proposals Through Decision-Maker Eyes with apex-review

Where this fits

You've just finished drafting a technical proposal—new architecture, a migration plan, or a conference talk abstract. Your agent validates syntax, checks facts, and even polishes prose. But the real question is: will the CTO approve this? Will the QA lead trust it? That's a different review layer, and it's exactly where apex-review steps in. Instead of asking "is this readable?", it asks "is this approvable?"—simulating the judgment of organizational decision-makers before you ever hit send.

Why agents benefit

• Persona-gated approval scoring: The skill evaluates your proposal from distinct roles (CTO, technical lead, QA lead, conference reviewer) and outputs a pass/fail per persona, so your agent can prioritize fixes by who's most likely to reject.
• HTML deck generation: It produces a presentation-ready deck summarizing the review, making it easy for your agent to share findings with stakeholders or archive them for future reference.
• Actionable improvement loop: It connects to sim-conductor for targeted suggestions, letting your agent iterate on the proposal until approval likelihood improves—not just grammar.
• Clear role separation: Unlike hub-persona-auditor (which checks reader comprehension), apex-review focuses on decision-making experience, avoiding redundant checks in your pipeline.
• Model-flexible execution: It works with Sonnet as a base and can route depth-critical steps to stronger models when needed, keeping cost and latency in check.

Practical scenarios

Pre-submission gate for internal architecture reviews

Your agent drafts a Kubernetes migration proposal. Before the architecture review board meets, apex-review runs it through the CTO and technical lead personas. It flags that the rollback plan is underspecified—a likely rejection point—and your agent revises it, turning a probable "no" into a confident "yes."

Conference talk abstract vetting

You're submitting to a major conference. The program committee persona in apex-review gives your abstract a low score because the learning outcomes aren't explicit. The skill suggests concrete rewording, and your agent applies it, increasing your acceptance odds.

Multi-stakeholder alignment for cross-team projects

A proposal affects data, QA, and operations. apex-review runs all three personas simultaneously, generating an HTML deck that shows where each lead would push back. Your agent then drafts targeted responses, smoothing the approval path before the actual meeting.

Add it to your agent workflow

Install the skill via npx:

npx skills add chrono-meta/forge-harness --skill apex-review

Then trigger it in your agent with /apex-review or ask "Will this work for the decision-maker?" The skill reads your proposal file, runs the persona reviews, and writes an HTML deck to your working directory. For a deeper look, see the skill page on OpenAgentSkill.

Compare before adopting

Before adding apex-review, check its GitHub activity (14 stars, but verify recent commits) and compare it with alternatives like hub-persona-auditor if you only need readability checks. Assess whether your workflow actually requires decision-maker simulation—if you're building internal tools, the CTO persona may be overkill. Also confirm the skill's model requirements align with your agent's runtime.

Why it is worth tracking

apex-review fills a real gap: most review skills focus on technical accuracy or clarity, not organizational approval dynamics. Its persona-based approach is novel and its integration with sim-conductor suggests a roadmap toward iterative proposal refinement. Even if you don't adopt it today, tracking it gives you a benchmark for how agent skills are evolving beyond surface-level checks—and when your proposal process needs that extra layer, you'll know exactly where to turn.