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/lugassawan/swe-workbenchWrote 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/commands/lugassawan/swe-workbench/test)<a href="https://agentmods.dev/commands/lugassawan/swe-workbench/test"><img src="https://agentmods.dev/badge/commands/lugassawan/swe-workbench/test/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/commands/lugassawan/swe-workbench/test"><img src="https://agentmods.dev/badge/commands/lugassawan/swe-workbench/test.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 | $0.00067 | $0.02140 |
| Opus 5.5 | $0.00027 | $0.00856 |
| Sonnet 5.5 | $0.00013 | $0.00428 |
| Haiku 4.5 | $0.00007 | $0.00214 |
Grade A, and why
test 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.
How it starts
The opening of the file, as written. The whole thing — 150 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Target: $ARGUMENTS
If $ARGUMENTS contains a ticket reference, invoke swe-workbench:ticket-context first and prepend its structured summary to the delegation context below. Skip if $ARGUMENTS is free-text with no recognizable ref. (Trigger patterns are defined in that skill's "When to invoke" section.)
Mode resolution
Parse $ARGUMENTS for --mode e2e-live, --mode e2e, or --e2e. Match --mode e2e-live by exact token first — a substring match on e2e would false-trip the e2e branch since e2e-live contains e2e:
- E2E-live mode (
--mode e2e-livepresent): strip the flag from the target and follow the E2E-live path below. - E2E mode (
--mode e2eor--e2epresent): strip the flag from the target and follow the E2E path below. This also means a bare--e2e-liveattempt (unsupported — only--mode e2e-liveis recognized) must not be treated as--e2e. - Default mode (no flag, or any unrecognized flag such as a bare
--e2e-live): follow the Unit path below with the unrecognized token left in the target text — there is no dedicated error path for unsupported flags.
E2E path (--mode e2e)
Hard gate
Before dispatching any agents, verify that Playwright MCP is available in this session by checking whether browser_snapshot (or equivalent browser_* tools under the MCP install prefix) is reachable.
If Playwright MCP is not connected, stop immediately and return:
BLOCKED: Playwright MCP not connected — run `claude mcp add playwright npx @playwright/mcp@latest`, restart Claude Code, and retry.
Do not delegate to any agent until the gate passes.
Pipeline
Dispatch the following two-agent pipeline in sequence:
swe-workbench:e2e-test-writer— explores the live app via Playwright MCP, authors durable spec files, and produces a behaviour inventory + spec paths.swe-workbench:e2e-test-verifier— receives the spec paths from step 1, runs them via the project's detected E2E command, and distrusts false-greens.
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 First seen · 150 lines · 67 tokens per session scan A 273984564738
test is a command published in the GitHub repository lugassawan/swe-workbench (3 stars, last pushed yesterday), licensed MIT. It adds 67 tokens to every session and 2,140 once invoked, about $0.0003 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 2026-08-31.
Other commands, from other repositories
recette
Tests d'acceptance automatises avec Claude in Chrome.
ia-test-browser
Run browser tests on pages affected by current PR or branch.
record
Record a browser walkthrough of a URL using Antigravity (agy). Generates .webm video, screenshots, and a report. Auto-converts to MP4 if ffmpeg is available.
qa
Smoke or browser-walk a running app. Report only. Do not implement. Do not merge.
record
Record the browser while validating a flow, then judge it against a rubric.
qa
Systematic QA testing with health scoring — tests web app, finds bugs, fixes them iteratively. Regression mode for re-testing known issues.