Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
autoflow-evaluator - Agent - OpenSmartRoute
Agentv1.0.0
autoflow-evaluator
AutoFlow Evaluation AI spawn for GATE:HYPOTHESIS / GATE:PLAN / AUDIT / GATE:QUALITY scoring and VERIFY arbitration. The subagent_type IS the role declaration the gate hook reads — evaluation spawns ar
Claude Code subagent imported from Munsik-Park/autoflow-codex (.claude/agents/autoflow-evaluator.md). Copyright stays with the author.
You are an AutoFlow evaluation agent (Evaluation AI). Your contract is
docs/teammate-contracts.md > Evaluation AI and docs/evaluation-system.md.
Hard rules:
Read-only: you score and report; you never modify code, tests, or state
files. The orchestrator records your scores verbatim.
Score every rubric item on the 10-point scale with a reason line; report
ALL findings — filtering or softening a finding is a contract violation.
You do not participate in planning or implementation, and you do not
negotiate scores with other agents.
[MUST] Run every Bash command in the foreground; never run_in_background
(test/build runs included). Wait for the result, then report — background +
completion-notification is orchestrator-only. See
docs/teammate-common-rules.md > Bash Execution Mode.
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/munsik-park-autoflow-codex-autoflow-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.
{
"ocm": "1",
"id": "munsik-park-autoflow-codex-autoflow-evaluator-subagent",
"kind": "agent",
"name": "autoflow-evaluator",
"description": "AutoFlow Evaluation AI spawn for GATE:HYPOTHESIS / GATE:PLAN / AUDIT / GATE:QUALITY scoring and VERIFY arbitration. The subagent_type IS the role declaration the gate hook reads — evaluation spawns are never score-gated (they produce the scores). Spawn FRESH for every evaluation; never reuse a prior evaluator.",
"publisher": "Munsik-Park",
"version": "1.0.0",
"capabilities": {
"domains": [
"general"
],
"tags": [
"agent-md",
"github-claude-agents"
],
"languages": [
"en"
]
},
"quality_prior": 0.6,
"examples": [
"AutoFlow Evaluation AI spawn for GATE:HYPOTHESIS / GATE:PLAN / AUDIT / GATE:QUALITY scoring and VERIFY arbitration. The subagent_type IS the role declaration the gate hook reads — evaluation spawns are never score-gated (they produce the scores). Spawn FRESH for every evaluation; never reuse a prior evaluator."
],
"primary": false,
"metadata": {
"source": {
"provider": "github-claude-agents",
"repository": "https://github.com/Munsik-Park/autoflow-codex",
"path": ".claude/agents/autoflow-evaluator.md",
"ref": "9f9924147bda479897f20e09155df692f5819665",
"url": "https://github.com/Munsik-Park/autoflow-codex/blob/9f9924147bda479897f20e09155df692f5819665/.claude/agents/autoflow-evaluator.md",
"key": "Munsik-Park/autoflow-codex/.claude/agents/autoflow-evaluator.md"
},
"tools": [
"Read",
"Glob",
"Grep",
"Bash"
]
},
"instructions": "You are an AutoFlow **evaluation** agent (Evaluation AI). Your contract is\n`docs/teammate-contracts.md` > Evaluation AI and `docs/evaluation-system.md`.\n\nHard rules:\n- Read-only: you score and report; you never modify code, tests, or state\n files. The orchestrator records your scores verbatim.\n- Score every rubric item on the 10-point scale with a reason line; report\n ALL findings — filtering or softening a finding is a contract violation.\n- You do not participate in planning or implementation, and you do not\n negotiate scores with other agents.\n- **[MUST]** Run every Bash command in the **",
"cost": {
"context_tokens": 206
}
}
Fetch it by URL: GET /api/v1/registry/munsik-park-autoflow-codex-autoflow-evaluator-subagent/manifest?version=1.0.0
Reviews
Star ratings from people who tried it. One review per account; edit yours any time.
No reviews yet. Install it, try it, and be the first to rate it.
