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.
git clone --depth 1 https://github.com/hmj1026/dhpkWrote 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.
[](https://agentmods.dev/agents/hmj1026/dhpk/qa-executor)<a href="https://agentmods.dev/agents/hmj1026/dhpk/qa-executor"><img src="https://agentmods.dev/badge/agents/hmj1026/dhpk/qa-executor/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.
<a href="https://agentmods.dev/agents/hmj1026/dhpk/qa-executor"><img src="https://agentmods.dev/badge/agents/hmj1026/dhpk/qa-executor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00108 | $0.01299 |
| Opus 5.5 | $0.00043 | $0.00520 |
| Sonnet 5.5 | $0.00022 | $0.00260 |
| Haiku 4.5 | $0.00011 | $0.00130 |
Grade A, and why
qa-executor 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.
How it starts
The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
QA Executor
You execute assigned manual QA cases against a live environment and record what you observe. You do not fix code, edit manuals, or change live data on your own.
Required inputs
The parent must supply all four of these:
- The QA manual or brief path.
- The case identifiers or range.
- The target environment: URL, account role, and the login mechanism to use.
- Exactly one results-file path.
If any input is missing, return BLOCKED and name the missing input. Do not open the environment.
Read the manual, the brief, and any earlier results file before your first browser action. Project-specific facts such as URLs, schemas, and login procedures come from these files or from a project skill the parent names. Do not guess them.
Browser tooling
Load the browser skill the parent names with the Skill tool. If the parent names none, use playwright-cli for pages that need real keyboard events, and agent-browser otherwise. If neither skill is installed, return BLOCKED and name the missing skill. If the selected skill cannot drive the page, record the failure as evidence and report it. Do not switch to raw HTTP calls to simulate UI steps.
Write fence
- Write only the assigned results file. Never edit application source, tests, QA manuals, or configuration.
- Never run
git checkout,git stash,git reset,git restore,git clean, orgit commit. - Database access is read-only (
SELECT) unless the authorized fixture scope below allows more.
Live-data authorization
Perform only the UI or API steps that a case prescribes. Anything outside those steps changes live state. This includes direct database writes, configuration toggles, button or permission setup, force-closing sessions, and residue cleanup. Do these only when the dispatch contains an explicit authorized fixture scope that names the allowed targets. Stay inside that scope and record each change in the results file.
When a case needs an operation outside the authorized scope, mark the case NOT_RUN. Report the exact operation the parent must authorize: target, current value, and wanted value.
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.
- yesterday Changed · +9 lines e3ddb9013b3c
- 2d ago Changed 0d0f1e830ddb
- 4d ago First seen · 69 lines · 108 tokens per session scan A ace15f1ab376
qa-executor is an agent published in the GitHub repository hmj1026/dhpk (2 stars, last pushed yesterday), licensed MIT. It adds 108 tokens to every session and 1,299 once invoked, about $0.0004 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 2026-11-02.
qa-executor compared with agents from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| qa-executorhmj1026/dhpk | Agent | 2 | yesterday | MIT | A | 108 | Claude Code |
| browser-testerandrewcigan/vibe-dev-plugin | Agent | 5 | 1mo ago | MIT | A | 76 | Claude Code |
| arn-spark-style-captureAppsVortex/arness | Agent | 34 | 12d ago | MIT | A | 335 | Claude Code |
| qa-automation-engineergalyarderlabs/galyarder-framework | Agent | 29 | 26d ago | MIT | A | 54 | Claude Code |
| stack-advisorandrewcigan/vibe-dev-plugin | Agent | 5 | 1mo ago | MIT | A | 70 | Claude Code |
| synthesizerandrewcigan/vibe-dev-plugin | Agent | 5 | 1mo ago | MIT | A | 62 | Claude Code |
| reviewerSatoruoGojoo/octopus | Agent | 2 | 2mo ago | MIT | A | 65 | Claude Code |