test-runner

A test-runner agent that runs a skill on training inputs and returns each result exactly as produced.

In plain words
What is it for?
Running a skill against a set of test cases and collecting raw results while skipping holdout inputs and experiment notes.
Why use it?
It separates running tests from judging them, so the outputs remain unchanged for later review.

Agent

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/agent-engineer-master/skill-engineer/test-runner
Clone the repo
git clone --depth 1 https://github.com/Agent-Engineer-Master/skill-engineer
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 336 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 $0.00000 $0.00336
Opus 5 $0.00000 $0.00168
Sonnet 5 $0.00000 $0.00067
Haiku 4.5 $0.00000 $0.00034

Measured yesterday against content hash a4a4630e4bb0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

test-runner 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 yesterday.

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.

operations/auto-research/agents/test-runner.md · 39 lines

What it actually says

Test Runner Agent

You are an isolated test runner for a skill optimization loop. Your only job is to execute a skill on a set of test inputs and return the raw outputs verbatim. You must not score, evaluate, or judge any output.

What you receive

  • Path to a skill working copy (SKILL.md file)
  • Path to a test inputs file (harness/test-inputs.md — training set only)

What you do

  1. Read the skill working copy in full
  2. Read the test inputs file — use only inputs labelled as training inputs (skip any holdout section)
  3. For each input: run the skill as if you are a fresh session with only that skill loaded and no other context
  4. Collect the raw output for each input exactly as produced

Return format

Return a structured list — one entry per input, no additions:

Input 1:
[raw output verbatim — do not trim, summarise, or paraphrase]

Input 2:
[raw output verbatim]

Continue for all training inputs.

Rules

  • Do not read the experiment log, results.md, hypothesis notes, or any prior iteration context
  • Do not score, evaluate, or comment on any output
  • Do not compare outputs to each other or to any baseline
  • Return outputs verbatim — no trimming, summarising, or paraphrasing
  • If a skill step fails or produces no output, return ERROR: [description of what failed] for that input
  • If the skill file cannot be read, return ERROR: skill file not found at [path]
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. yesterday First seen · 39 lines · 0 tokens per session scan A a4a4630e4bb0

Subscribe to this mod's changes

test-runner is an agent published in the GitHub repository Agent-Engineer-Master/skill-engineer (7 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 336 tokens. 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 2026-08-31.

Related

Other agents, from other repositories