AgentiaPT/vela-slides / .claude/skills/burst-bug-hunter
burst-bug-hunter
VerifiedTime-boxed adversarial bug-hunting of a LIVE app driven by MULTI-STEP PLAYWRIGHT BURSTS against a persistent warm browser — not one LLM turn per CLI step. The app is opened ONCE; the agent submits full scripts that run to completion unattended (one structured result per burst); reset to a known initial state between scenarios; inherit the implementer's driver verbs instead of re-predicting selectors. Generic / repo-agnostic: all app specifics come from the repo's own config. Use when hunting bugs in a browser-drivable app under a hard wall-clock budget.
| File | Kind | Size |
|---|---|---|
| .gitignore | other | 20 B |
| SKILL.md | skill md | 5.9 KB |
| assets/driver-server.mjs | script | 20.4 KB |
| assets/example-burst.mjs | script | 1.3 KB |
| assets/start-hunt.shexecutable | script | 744 B |
| assets/test-timeout-isolation.mjs | script | 18.9 KB |
| assets/vrunexecutable | script | 2.0 KB |
assets/example-burst.mjs
View on GitHub// example-burst.mjs — TEMPLATE for a defensive multi-step burst. Copy + adapt.
// Import the REPO's verb library (path from the repo's .hyper-sprint config), not any
// verbs from this generic skill. A burst runs unattended to completion; make it record
// a per-step trace so a wrong prediction fails at a NAMED step (cheap to correct) instead
// of silently derailing.
//
// import * as V from "file:///abs/path/to/<repo>/.hyper-sprint/<verbs>.mjs";
export async function run(page, ctx) {
const trace = [];
const step = async (name, fn) => {
const t0 = Date.now();
try { const r = await fn(); trace.push({ step: name, ok: true, ms: Date.now() - t0, r: r ?? null }); return r; }
catch (e) { trace.push({ step: name, ok: false, ms: Date.now() - t0, err: String(e.message || e) }); throw e; }
};
const out = { trace };
try {
// await step("do X", () => V.someVerb(page, ...));
// out.observed = await step("read state", () => page.evaluate(() => /* inspect DOM */));
// ctx.reset() / ctx.shot("name") available. Use waitForFunction/waitForSelector, NOT fixed sleeps.
out.reachedEnd = true;
} catch (e) {
out.failedAt = trace[trace.length - 1]; // exactly which predicted step was wrong
out.reachedEnd = false;
}
return out;
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |