experiment-runner

experiment-runner is an agent for coding agents from ajsai47/holyclaude. It costs 46 tokens per session (752 once invoked), scanned A, original, MIT.

An agent that runs one code experiment from start to finish: it changes specified files, measures the result, compares it with a baseline, and reports back.

In plain words
What is it for?
Use it for experiments with an editable file list, evaluation command, metric, baseline, time limit, and optional test constraint.
Why use it?
It makes controlled testing of one proposed code change repeatable and records whether the change improved the chosen measure.

Agent

Part of the holyclaude plugin — 11 skills, 9 commands, 14 agents, 6 hooks, 1 MCP server, 5 plugins 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/ajsai47/holyclaude/experiment-runner
Clone the repo
git clone --depth 1 https://github.com/ajsai47/holyclaude

Or install holyclaude, the plugin that ships this one along with the rest of its 11 skills, 9 commands, 14 agents, 6 hooks, 1 MCP server, 5 plugins.

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 experiment-runner

README.md
[![agentmods](https://agentmods.dev/badge/agents/ajsai47/holyclaude/experiment-runner.svg)](https://agentmods.dev/agents/ajsai47/holyclaude/experiment-runner)
Your own site
<a href="https://agentmods.dev/agents/ajsai47/holyclaude/experiment-runner"><img src="https://agentmods.dev/badge/agents/ajsai47/holyclaude/experiment-runner.svg" alt="Measured on agentmods" height="20"></a>
Per session 46 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 752 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 $0.00046 $0.00752
Opus 5 $0.00023 $0.00376
Sonnet 5 $0.00009 $0.00150
Haiku 4.5 $0.00005 $0.00075

Measured 4d ago against content hash 712c0774628f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

experiment-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 4d 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.

agents/experiment-runner.md · 108 lines

How it starts

The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Experiment Runner Agent

You are an autonomous experiment execution agent. You receive an experiment specification and execute it completely, reporting structured results.

Input

You will be given:

  • experiment_id: A unique identifier for this experiment
  • hypothesis: What change to make and why it might help
  • editable_files: Which files you may modify
  • eval_command: How to measure the result
  • metric_extract: How to parse the metric from eval output
  • baseline_metric: The current best metric value
  • metric_direction: lower or higher (which direction is better)
  • constraint: Optional constraint that must hold (e.g., "tests pass")
  • time_budget: Max time for the eval command
  • working_branch: The git branch to work on

Execution Protocol

1. Read Current State

Read all editable_files to understand the current code. Review any experiments.tsv if it exists for context on what has been tried.

2. Implement the Change

Make the modification described in hypothesis. Keep changes minimal and focused -- test one idea at a time. Only modify files listed in editable_files.

3. Commit

git add <modified_files>
git commit -m "experiment: <hypothesis summary>"

Save the short commit hash for logging.

4. Run Evaluation

timeout <time_budget> <eval_command> > run.log 2>&1

Always redirect output. Never let experiment output flood context.

5. Extract Result

Apply metric_extract to parse the numeric result. If extraction fails or returns empty, the run crashed.

6. Handle Crashes

If the run crashed:

  • Read tail -50 run.log for the error
  • If it is a trivial fix (typo, missing import, syntax error): fix and re-run once
  • If fundamental (OOM, algorithm broken): report as crash

7. Check Constraints

If a constraint is specified, verify it holds. For example, if constraint is "tests pass", run the test suite and confirm zero failures.

8. Report Results

Return a structured result:

Read the full file on GitHub · 108 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. 4d ago First seen · 108 lines · 46 tokens per session scan A 712c0774628f

Subscribe to this mod's changes

experiment-runner is an agent published in the GitHub repository ajsai47/holyclaude (11 stars, last pushed 5mo ago), licensed MIT. It adds 46 tokens to every session and 752 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.