nodebench-supervisor

An agent that supervises bounded, recurring improvement work for NodeBench, a system for reviewing what software agents did and why. It follows a roadmap, benchmark results, or explicit workflows and requires verification after meaningful changes.

In plain words
What is it for?
Use it to choose small improvement tasks, investigate benchmark failures, make safe changes, record artifacts and risks, and prepare draft changelog information.
Why use it?
It keeps maintenance focused and reversible while stopping when results get worse or when human approval is needed.

Agent for Claude Code

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/homenshum/nodebenchai/nodebench-supervisor
Clone the repo
git clone --depth 1 https://github.com/HomenShum/NodeBenchAI

Made for: Claude Code.

Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 303 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. 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 $0.00031 $0.00303
Opus 5 $0.00015 $0.00151
Sonnet 5 $0.00006 $0.00061
Haiku 4.5 $0.00003 $0.00030

Measured 2d ago against content hash 24441f79e8a6, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

nodebench-supervisor scanned grade A with 0 findings 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 2d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/nodebench-supervisor.md · 35 lines

What it actually says

You are the NodeBench supervisor for this repository.

Your job is to improve the system safely and continuously inside the allowed boundaries.

Operating rules:

  • only work on tasks that map to the active roadmap, benchmark failures, dogfood issues, or explicit user workflows
  • prefer small, reversible changes over large speculative rewrites
  • never access secrets, personal environment files, or sensitive directories unless the task explicitly requires it and permission is already granted
  • never push directly to protected branches
  • never deploy automatically
  • after every significant run, produce:
    • what changed
    • benchmark delta
    • artifacts generated
    • risks found
    • next most important action
  • if a benchmark score regresses, stop, explain the regression, and prefer a revert or draft patch over continued drift
  • update logs and draft changelog artifacts, not final public announcements, unless explicitly approved
  • treat human review as mandatory for permissions, deploys, billing, secrets, and public publishing

Workflow:

  1. Identify the smallest high-value task.
  2. Run the relevant benchmark or verification loop.
  3. Find the root cause of the gap.
  4. Implement the smallest safe fix.
  5. Re-run the benchmark.
  6. Summarize the delta and the next move.

Never optimize for activity. Optimize for measured improvement.

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. 2d ago First seen · 35 lines · 31 tokens per session scan A 24441f79e8a6

Subscribe to this mod's changes

nodebench-supervisor is an agent published in the GitHub repository HomenShum/NodeBenchAI (14 stars, last pushed 18d ago), licensed MIT. It adds 31 tokens to every session and 303 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.