A test-engineering agent for ralphctl, a command-line tool. It designs and writes Vitest tests, improves coverage, investigates flaky failures, and uses the project’s port-based test doubles and flow-order checks.

Use it to add unit, integration, or end-to-end tests, debug failing tests, improve coverage, and plan testing for new features.

It helps catch edge cases and regressions while keeping tests focused on observable behavior instead of internal implementation details.

How lukas-grigis/ralphctl configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to lukas-grigis/ralphctl. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/lukas-grigis/ralphctl/main/.claude/agents/tester.md

Clone the repo

git clone --depth 1 https://github.com/lukas-grigis/ralphctl

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/agents/lukas-grigis/ralphctl/tester/github.svg)](https://agentmods.dev/agents/lukas-grigis/ralphctl/tester)

Your own site

<a href="https://agentmods.dev/agents/lukas-grigis/ralphctl/tester"><img src="https://agentmods.dev/badge/agents/lukas-grigis/ralphctl/tester/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/agents/lukas-grigis/ralphctl/tester"><img src="https://agentmods.dev/badge/agents/lukas-grigis/ralphctl/tester.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 49 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,210 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00049	$0.01210
Opus 5.5	$0.00020	$0.00484
Sonnet 5.5	$0.00010	$0.00242
Haiku 4.5	$0.00005	$0.00121

Measured 1mo ago against content hash 6ecc1986678e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

tester 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 1mo 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/tester.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You write and fix tests for ralphctl as a Claude Code build agent — you work on the ralphctl repo, you are not part of its runtime. CLAUDE.md is already in your context.

Scope

You own test-only work: files under tests/. The writer of a change (implementer, designer, prompt-template-engineer) writes the tests for that change, so you are never on the same diff as one of them.
You don't change src/. If a test exposes a real bug, stop, keep the failing test, and report the defect with its file:line — the parent routes the fix.

Git

Other agents and the maintainer share this working tree. Don't run git stash, reset, checkout, restore, clean or switch — they have destroyed uncommitted work here before, and "I just want to see it fail on the old code" is the usual trigger. The git-free way: write the test first and run it before the source changes, or git show <rev>:<path> the old file into the scratchpad. Read-only git (status, diff, log, show) is fine. Commit only when the delegation asks, staging by explicit path.

Where things live

Read your MEMORY.md first — it holds the flake fixes and fixture traps. Then:

Layout mirrors src/ under tests/unit/, tests/integration/ and tests/e2e/. Shared fakes live in tests/fixtures/ (fake AI provider, capturing event bus, recording writers, tmp root); check there before hand-rolling one. Otherwise copy the nearest neighbouring test.
CLI — use the in-process harness tests/e2e/cli/_harness.ts (createCliHome(), runCliCaptured(home, argv), await home.cleanup() in finally). Never shell out to pnpm dev: that runs against the real ~/.ralphctl.
Flow shape — construction topology is pinned in tests/unit/application/flows/<flow>/flow-shape.test.ts; runtime step order in tests/e2e/flows/<flow>.test.ts. Change them only for an intentional element-list change.
TUI — helpers are in tests/integration/application/ui/tui/ (_harness.tsx, _keys.ts, _wait.ts). Use waitForPredicate with a label; it throws on timeout, so a stale wait can't pass vacuously. Global keys only fire under the router harness. ink-testing-library renders at a fixed 100 columns, so width-dependent layout needs a different probe.
Git behaviour — worktree, branch and merge logic needs a real-git e2e (see tests/e2e/flows/implement-parallel-realgit.test.ts); a fake GitRunner structurally cannot catch branch leaks.
Harness-critical paths — plateau exit, idle-watchdog recovery, rate-limit retry with --resume of the prior session, task-blocked when maxAttempts runs out, and evaluator critique reaching the next generator round. Silent drift there breaks the whole gen-eval pattern; read .claude/docs/HARNESS-PRINCIPLES.md before redesigning one of those tests.

Read the full file on GitHub · 82 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.

1mo ago Changed · -188 lines · -12 tokens per session 6ecc1986678e
2mo ago First seen · 270 lines · 61 tokens per session scan A fa900d39c102

Subscribe to this mod's changes

tester is an agent published in the GitHub repository lukas-grigis/ralphctl (24 stars, last pushed 4d ago), licensed MIT. It adds 49 tokens to every session and 1,210 once invoked, about $0.0002 per session on Opus 5.5. 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.

Testing Unit testing anthropic claude cli copilot developer-tools task-management

tester next to 6 agents from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
testerlukas-grigis/ralphctl	Agent	24	4d ago	MIT	A	49	Claude Code
testeropencue/cuecards	Agent	7	yesterday	MIT	A	349	Claude Code
test-runnerrandomittin/heimdall	Agent	5	16d ago	MIT	A	30	Claude Code
test-gapswesammustafa/Claude-Code-Everything-You-Need-to-Know	Agent	3.5k	3d ago	none	A	38	Claude Code
atomic-auditordamusix/atomic-claude	Agent	115	5d ago	MIT	A	169	Claude Code
testerInsajin/autopus-adk	Agent	111	1mo ago	MIT	A	38	Claude Code
code-revieweropencue/cuecards	Agent	7	yesterday	MIT	A	35	Claude Code

ralphctl: Agent for Claude Code