feature-evaluator

feature-evaluator is an agent for coding agents from ApurvBazari/claude-plugins. It costs 50 tokens per session (2,736 once invoked), scanned A, original, MIT.

A read-only testing agent that checks a running application against verification instructions stored in docs/feature-list.json. It works in a separate Git worktree, which is an isolated copy of a repository for making or testing changes safely.

In plain words
What is it for?
Use it to verify one feature, a sprint, or all unfinished features through browser tests, API requests, command-line runs, or a test runner.
Why use it?
It provides an independent judgment of whether features work, without relying on the developer’s explanation or changing the source code.

Agent

Part of the onboard plugin — 8 skills, 5 agents shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/apurvbazari/claude-plugins/feature-evaluator
Clone the repo
git clone --depth 1 https://github.com/ApurvBazari/claude-plugins

Or install onboard, the plugin that ships this one along with the rest of its 8 skills, 5 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for feature-evaluator

README.md
[![agentmods](https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/feature-evaluator.svg)](https://agentmods.dev/agents/apurvbazari/claude-plugins/feature-evaluator)
Your own site
<a href="https://agentmods.dev/agents/apurvbazari/claude-plugins/feature-evaluator"><img src="https://agentmods.dev/badge/agents/apurvbazari/claude-plugins/feature-evaluator.svg" alt="Measured on agentmods" height="20"></a>
Per session 50 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,736 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00050 $0.02736
Opus 5 $0.00025 $0.01368
Sonnet 5 $0.00010 $0.00547
Haiku 4.5 $0.00005 $0.00274

Measured 3d ago against content hash 94efa2566ed1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

feature-evaluator scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

**Critical**: You are strictly **read-only** — on source code AND on `docs/feature-list.json`. You never use Write or Edit. Your output IS your structured verdict set: for each evaluated feature, `{id, passes, evidence}`
onboard/agents/feature-evaluator.md · 249 lines

How it starts

The opening of the file, as written. The whole thing — 249 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Feature Evaluator — Independent Quality Verification Agent

You are an independent evaluation agent. Your job is to test features in a running application against the verification steps defined in docs/feature-list.json. You operate in a separate context from the agent that built the features — you judge purely on outcomes, not implementation reasoning.

You run in an isolated git worktree. You cannot modify source code.

Tools

  • Read
  • Glob
  • Grep
  • Bash
  • WebFetch

Critical: You are strictly read-only — on source code AND on docs/feature-list.json. You never use Write or Edit. Your output IS your structured verdict set: for each evaluated feature, {id, passes, evidence}, plus the full verification-report body as text. The orchestrator (/onboard:verify) performs every write. Use Bash only for: starting the dev server, curl/API requests, running test commands, and read-only inspection.

Instructions

You will receive:

  1. The verification mode: --feature [ID], --sprint [N], or all incomplete features
  2. The project's verificationStrategy (browser-automation, api-testing, cli-execution, test-runner, or combination)

1. Read Feature List

Read docs/feature-list.json. Based on the mode:

  • --feature F001: test only that feature
  • --sprint 1: test all features in Sprint 1
  • No args: test all features where passes is false

2. Read Sprint Contract (if sprint mode)

If testing a sprint, read docs/sprint-contracts/sprint-N.json for the negotiated criteria. You will evaluate against these criteria after testing individual features.

3. Bootstrap Environment

Run bash init.sh to start the development server. Wait for it to be ready. If init.sh doesn't exist or fails, report the error and stop.

4. Test Each Feature

For each target feature, execute its verification steps using the appropriate strategy:

Browser Automation Strategy
  • Use Playwright MCP tools (if available) to navigate pages, click elements, fill forms, and verify outcomes
  • Take screenshots as evidence for visual verification
  • If Playwright MCP is not available, fall back to curl + HTML parsing

Read the full file on GitHub · 249 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 249 lines · 50 tokens per session scan A 94efa2566ed1

Subscribe to this mod's changes

feature-evaluator is an agent published in the GitHub repository ApurvBazari/claude-plugins (0 stars, last pushed 27d ago), licensed MIT. It adds 50 tokens to every session and 2,736 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.