'Do any of you actually test for prompt injection?' — devs admit shipping LLM apps untested
Dangerous_Net517 · reddit · 2026-10-18
A Reddit developer admits having almost no real process for testing LLM apps against prompt injection, system prompt leaks, or misuse once models are wired to tools and data — mostly eyeballing and hoping. The post asks whether others actually test before shipping (manually, via scripts, or with tools), what risks they worry about most, whether anyone has had a real production incident, and whether non-testers skip it because their use case is low-risk or simply because no easy method exists. The thread surfaces an industry-wide gap: security testing for LLM apps remains ad hoc, with most teams 'shipping and hoping.'
More from coding & agent
- Teknium confirms Hermes Agent will support Android — Teknium · 2026-10-18
- DavidKPiano: I appreciate how Claude Code pauses to ask me questions — DavidKPiano · 2026-10-18
- Pantheon brings 18 Greek mythology themes to Nous Research's Hermes Desktop in one plugin — Teknium · 2026-10-18
- User dubs English movies to Polish locally with Hermes Agent + OpenVoice on a 16GB M4 Mac mini — NousResearch · 2026-10-18
- Long chats forget after summarizing, so this dev built a memory MCP server — Jolly-Scientist-2484 · 2026-10-18
- Link agent wallet v0.27.5 expands agent payments beyond one-time e-commerce purchases — jeff_weinstein · 2026-10-18