# @maestria/prime-agent

<div class="plugin-root">
  <div class="home-glow" aria-hidden="true"></div>
  <div class="home-grid" aria-hidden="true"></div>

  <div class="home-inner">
    <PlatformHero
      eyebrow="@MAESTRIA/PRIME-AGENT"
      title="maestria skills for Prime Agent."
      lead="Deliver the maestria engineering methodology to Prime Agent as standard Agent Skills, plus a small verified extension for workflow-mode commands and mode prompt injection."
      mark={marks.primeAgent}
      primaryCta={{ href: '/prime-agent/getting-started/installation/', label: 'Get started' }}
      secondaryCta={{ href: '/prime-agent/getting-started/quick-start/', label: 'Quick start' }}
      installCommand="npx maestria install prime-agent"
    />

    <PluginSection
      index="01"
      kicker="WHAT YOU GET"
      title="14 skills and one verified extension."
      class="pt-[clamp(3.5rem,7vw,5rem)] mb-[clamp(4rem,9vw,7rem)]"
    >
      <FeatureGrid
        items={[
          { title: 'Full specialist roster', desc: 'Seven specialists: adventurer, architect, builder, diagnose, planner, reviewer, writer.' },
          { title: 'Orchestration and rules', desc: 'The orchestrator, global-rules contract, handoff aid, and iteration-limits aid ship as skills.' },
          { title: 'Workflow modes', desc: 'fein, sonar, and blitz skills mirror the core pipeline stages.' },
          { title: 'Extension: mode commands', desc: 'dist/extension.mjs registers /fein, /sonar, /blitz, /mode-clear, and /maestria-status.' },
          { title: 'Extension: mode injection', desc: 'Injects the active mode prompt on every agent turn.' },
          { title: 'Spec-shaped skills', desc: 'Every skill is a standard Agent Skill at skills/<name>/SKILL.md with the required name/description frontmatter.' },
        ]}
      />
    </PluginSection>

    <PluginSection
      index="02"
      kicker="VERIFICATION STATUS"
      title="Native candidate - verified subset only."
      class="mb-[clamp(4rem,9vw,7rem)]"
    >
**Runtime behavior is not yet tested end to end:** The generated skills match the documented [Prime Agent](https://github.com/PrimeIntellect-ai/prime-agent)
        Agent Skills contract and the extension is verified against the pinned Prime fork's public extension API,
        but runtime behavior is **not yet tested end to end** in a live Prime session. Native recursive-subagent
        (`rlm`) dispatch and JSON/RPC headless-mode integration remain **deferred**.
      <FeatureGrid
        columns={2}
        items={[
          { title: 'Verified subset only', desc: "The extension covers only mode commands and mode prompt injection. Prime's rlm(...) subagent dispatch is an IPython-side tool with no public JS extension bridge, so this package cannot spawn child agents." },
          { title: 'Advisory, not enforced', desc: 'Prime Agent has no skill-level tool-denial mechanism, so read-only roles describe their intent without claiming enforcement.' },
          { title: 'Not a sandbox', desc: 'Prime Agent executes model-generated Python and project commands with your user permissions. Restrict use to trusted repositories, skills, and instructions.' },
        ]}
      />
    </PluginSection>

    <RelatedDocs
      links={[
        { label: 'Specialist reference', href: '/core/agents/' },
        { label: 'Workflow patterns', href: '/core/workflow-patterns/' },
        { label: 'Changelog', href: '/prime-agent/changelog/' },
        { label: 'Contributing', href: '/prime-agent/contributing/' },
      ]}
    />

    <SiteFooter />

  </div>
</div>