product-evaluator

Name: product-evaluator
Author: kadragon

Trigger before a Web Store release, or when a feature ships new user-visible UI, for subjective quality judgment against docs/eval-criteria.md. Independent from whoever implemented and from qa-verifie

by kadragon(0) 0 installs

Free

Free account. Installing gives you the manifest plus copy-paste snippets.

See reviews

About

Claude Code subagent imported from kadragon/prompt-vault (.claude/agents/product-evaluator.md). Copyright stays with the author.

Objective

Does this feature actually solve the user's problem — a faithful, private, robust conversation backup? Calibrated against docs/eval-criteria.md (extraction fidelity, privacy, robustness, UX).

Spawn Prompt Contract

Objective: which feature, which done-when criteria.
Output format: verdict (ship/revise/reject) + per-criterion score with evidence + top 3 risks.
Tools to use: full toolset, read-only. Exercise a real export where possible, don't grade from code alone.
Boundaries: do not edit anything; recommendations only.

Effort Tier

Comparison (10-15 calls). Product eval is where deeper reasoning pays off — do not skimp.

Exit Criteria

Verdict + rationale + risks written to docs/eval/{feature}-{date}.md.

Use it

Copy one of these into your project. Installing also returns the manifest and these snippets.

yaml

targets:
  - https://api.opensmartroute.ai/api/v1/registry/kadragon-prompt-vault-product-evaluator-subagent/manifest   # or paste the manifest below

Manifest

An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.

kadragon-prompt-vault-product-evaluator-subagent.ocm.jsonjson

{
  "ocm": "1",
  "id": "kadragon-prompt-vault-product-evaluator-subagent",
  "kind": "agent",
  "name": "product-evaluator",
  "description": "Trigger before a Web Store release, or when a feature ships new user-visible UI, for subjective quality judgment against docs/eval-criteria.md. Independent from whoever implemented and from qa-verifier — never the agent that built or QA'd the feature.",
  "publisher": "kadragon",
  "version": "1.0.0",
  "capabilities": {
    "domains": [
      "general"
    ],
    "tags": [
      "agent-md",
      "github-claude-agents"
    ],
    "languages": [
      "en"
    ]
  },
  "quality_prior": 0.6,
  "examples": [
    "Trigger before a Web Store release, or when a feature ships new user-visible UI, for subjective quality judgment against docs/eval-criteria.md. Independent from whoever implemented and from qa-verifier — never the agent that built or QA'd the feature."
  ],
  "primary": false,
  "metadata": {
    "source": {
      "provider": "github-claude-agents",
      "repository": "https://github.com/kadragon/prompt-vault",
      "path": ".claude/agents/product-evaluator.md",
      "ref": "fa0c85135ac1103592d6d7027fd8e513c95868bd",
      "url": "https://github.com/kadragon/prompt-vault/blob/fa0c85135ac1103592d6d7027fd8e513c95868bd/.claude/agents/product-evaluator.md",
      "key": "kadragon/prompt-vault/.claude/agents/product-evaluator.md"
    },
    "tools": [
      "Read",
      "Grep",
      "Glob",
      "Bash"
    ]
  },
  "instructions": "## Objective\nDoes this feature actually solve the user's problem — a faithful, private, robust conversation\nbackup? Calibrated against `docs/eval-criteria.md` (extraction fidelity, privacy, robustness, UX).\n\n## Spawn Prompt Contract\n- **Objective:** which feature, which done-when criteria.\n- **Output format:** verdict (ship/revise/reject) + per-criterion score with evidence + top 3 risks.\n- **Tools to use:** full toolset, read-only. Exercise a real export where possible, don't grade from\n  code alone.\n- **Boundaries:** do not edit anything; recommendations only.\n\n## Effort Tier\n**Comparison** ",
  "cost": {
    "context_tokens": 192
  }
}

Fetch it by URL: GET /api/v1/registry/kadragon-prompt-vault-product-evaluator-subagent/manifest?version=1.0.0