fault-injection

🏛️ ACM EASE '26 / VibeX full paper · DOI 10.1145/3816483.3816631 · Adversarial agent loops for verifiable vibe researching. Falsification-first research runtime.

Deterministic fault injection and autonomous resilience tests for DeepSeek Harness tools

Local dev proxy. Named services, HTTPS, and failure injection — no ports, no mocks.

Debug and fault-test any WebRTC session — Chrome extension, MCP server for AI coding agents, npm library, and Playwright plugin. Live RTCPeerConnection/WebSocket inspection, reconnect testing, network-loss simulation.

Distributed Mutual Exclusion Explorer (Variant 3): interactive Token Ring & Ricart–Agrawala visualiser with fault injection, scripted demos, and evidence export.

Deterministic 8-node chaos and soak testing for causal ordering, deduplication, replay, and recovery behavior.

Reproducible teardown of the openai-agents-js financial-research example: the manager emits a research report when every web search failed and when verification never passed. 6-case synthetic-orchestration harness, baseline 4/6, one-file patch 6/6, plus a field guide for probing your own agent.

Card & dice calculators for a bar table: liar's dice, hold'em, blackjack, 24 game. The substance isn't the features — it's the evidence the math is right: 395 assertions, 78 fault injections, eight rounds of adversarial review.

Deterministic local HTTP fault-injection proxy for testing retries, timeouts, and idempotency.

Fault-injection experiments for tool-using agents under partial failure

Agent Skill for Claude Code, OpenAI Codex, Cursor, and AI coding agents that catches duplicate side effects across retries, webhooks, jobs, concurrency, restarts, and migrations.

Reproducible teardown of a scheduled maintenance agent in google/adk-python: every issue audit fails, the batch reports them all as successfully processed, and the run exits 0. Harness, one-file fail-closed patch, hashed RED to GREEN evidence. Reported upstream first.