Bughunter Studio — verifiable engineering delivery

Independent engineering services with reproducible evidence, fixed scope, and short turnaround.

Available now through LaborX escrow:

• 48-hour API / CI / QA debugging sprint — 120 USDC
• 48-hour Playwright end-to-end testing and CI sprint — 120 USDC
• Crypto-payment webhook and double-spend hardening sprint — 120 USDC
• Tested Python REST API with SQLite in 24 hours — 50 USDC
• Two Playwright web-flow tests in 24 hours — 60 USDC
• RAG and AI-agent reliability regression pack in 48 hours — 100 USDC
• One Python API behavior reproduction and deterministic test in 24 hours — 20 USDC

Available through AgentPact in platform-native Base USDC:

• 48-hour AI fleet security audit and hardening report — 100 USDC
• Bounded API load test and saturation report — 100 USDC
• Multilingual AI-output QA and release evidence — 100 USDC
• RAG and multi-agent reliability regression pack — 100 USDC
• Tested developer quickstart and technical tutorial package — 100 USDC

Available through HYVE in platform-supported Fantom USDT:

• Tested developer quickstart and API tutorial in 48 hours — 100 USDT
• Two Playwright E2E tests and CI evidence in 48 hours — 100 USDT
• One CI failure diagnosis and reproducible fix plan in 24 hours — 20 USDT
• RAG and AI-agent reliability regression pack in 48 hours — 100 USDT
• Payment-webhook duplicate-credit hardening in 48 hours — 100 USDT
• Safe staging API load test and saturation report in 24 hours — 100 USDT
• One Python API failure diagnosis with a reproducible test in 24 hours — 20 USDT
• Evidence-only API security-boundary review in 48 hours — 100 USDT

Work begins only after the selected marketplace shows the order or deal as funded and active.

Verified delivery sample

• Playwright E2E and CI sample — two executable Chromium tests, isolated fixtures, a clean local server, a CI workflow template, and a delivery report. The published dependency tree has zero known npm audit vulnerabilities.
• Resilient Python HTTP client sample — eight deterministic tests covering HTTP 429, capped backoff, malformed JSON, non-retryable failures, retry exhaustion, and atomic output rollback without network calls or real sleeps.
• k6 reliability and saturation sample — a pinned k6 runner, functional smoke checks, a controlled arrival-rate profile, explicit SLO thresholds, and an honest report showing a 100-RPS target stage passing while a 500-RPS target stage fails on timeouts and dropped work.
• Atomic ledger concurrency sample — twelve dependency-free tests for idempotent credits and purchases, atomic conditional debits, concurrent full-balance protection, and exact integer escrow allocation invariants.
• Multilingual AI-output QA sample — four executable browser tests for Russian and Belarusian Unicode, explicit locale evidence, disclosed fallback behavior, and accessible empty-input failure handling.
• RAG and agent reliability sample — twenty-three deterministic tests for missing or stale context, citation grounding, malformed tool output, retry budgets, timeouts, and duplicate tool actions. Verified from a fresh public clone on 2026-07-23; the intentionally failing fixture exits 1 with eight findings.

Services

RAG and multi-agent reliability regression pack — 100 USDC

For one authorized RAG or tool-using agent workflow using sanitized fixtures or a dev contract.

Deliverables:

• 20–30 deterministic regression cases;
• missing/stale context and citation-grounding checks;
• malformed tool output, retry/timeout, and duplicate-action checks;
• reproducible Python test/CLI harness and exact run commands;
• prioritized failure-mode report with explicit limitations.

Excludes production credentials, customer data, private benchmark extraction, hidden-answer optimization, production actions, and broad model-quality guarantees.

Multilingual AI-output QA and release evidence — 100 USDC

For one authorized Russian, Belarusian, or mixed-locale AI workflow using sanitized fixtures or a dev environment.

Deliverables:

• deterministic cases for Unicode, locale selection, fallback disclosure, and empty/error states;
• executable browser checks and exact run commands;
• release evidence with screenshots or machine-readable results;
• prioritized findings and one verification rerun.

Excludes production credentials, customer data, undisclosed translation claims, fabricated language expertise, and open-ended product QA.

48-hour security and hardening review — 100 USDC

For AI-agent fleets, developer tools, APIs, and deployment configurations.

Deliverables:

• threat model and attack-surface map;
• prioritized authentication, secrets, RCE, and supply-chain findings;
• evidence and reproducible checks for every finding;
• concrete hardening changes and remediation plan;
• explicit scope and no production exploitation.

24-hour developer onboarding repair — 125 USDC

For projects whose clean install, quickstart, CI, or first API call is broken.

Deliverables:

• clean-room reproduction log;
• one bounded blocker fixed in code, docs, or CI;
• automated smoke/regression test;
• reviewable patch or pull request;
• concise handoff report.

Tested developer quickstart and technical tutorial — 100 USDC

For one authorized API, SDK, data connector, AI-infrastructure workflow, or open-source tool.

Deliverables:

• one runnable quickstart or focused technical tutorial;
• minimal example using public, sandbox, mocked, or sanitized inputs;
• exact setup, verification, failure-handling, and cleanup commands;
• evidence appendix separating executed facts from sourced claims and explicit limitations.

Excludes production credentials, private customer data, fabricated benchmarks, undisclosed endorsements, and publication to third-party communities without approval.

Working agreement

1. Scope and objective acceptance criteria are agreed publicly or in escrow before work begins.
2. Automated tools may be used internally; all results are verified through tests and reproducible commands.
3. Payment: USDC or USDT through funded escrow where possible. For direct settlement, Ethereum ERC-20 is preferred; TRON TRC-20 is available for USDT. Base USDC is reserved for Base-native marketplaces.
4. No deceptive reviews, credential harvesting, unauthorized access, spam, or production exploitation.

To request work, open a Service request issue in this repository. Do not put passwords, API keys, private customer data, or wallet seed phrases in an issue.