SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Travel Site Uses LLM for Moderation but Lets Humans Have the Final Say on Negative Reviews

0
·3 views

A developer building 'Back From My Trip', a travel review site, has designed an AI moderation pipeline that explicitly protects negative reviews from being automatically rejected. The system assigns every piece of content a moderation status — approved, needs_review, or rejected — with row-level database security ensuring only approved content is publicly visible. While the AI can approve content autonomously, it cannot issue a final rejection for text-based opinions; any AI-flagged rejection is routed to a human admin queue for confirmation or override. When the moderation system fails or exceeds its budget, content defaults to needs_review rather than auto-approval, ensuring failures never allow spam through or silence a legitimate opinion. The developer also patched an early security flaw where client-side text could be swapped before moderation, replacing it with a server-side trigger system that processes only stored row IDs.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Technical English for Security Engineers: A Practical Guide to Workplace Communication

A Russian-language guide published on DEV Community addresses the challenge many Russian-speaking IT and security professionals face when using English in international work environments. The article argues that the goal is not to master the entire language but to build a focused toolkit of phrases and patterns for specific workplace scenarios. Key use cases covered include writing status updates, explaining security findings, describing risk and impact, and communicating professionally in tools like Slack. The guide emphasizes that effective technical English prioritizes clarity and actionability over linguistic complexity, stressing that messages should answer what is happening, who owns the next step, and what action follows. The author draws on personal experience working with international engineering teams and American recruiters to illustrate how practical communication skills matter more than formal language proficiency.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Ephemora Cell WASM sandbox blocks all 8 attack vectors that Docker missed for AI code

Ephemora Cell is an open-source, capability-based WebAssembly sandbox designed to safely execute untrusted code written by AI agents and tools. Unlike traditional container approaches, it operates beneath existing agent frameworks, granting guest code only explicitly permitted capabilities while blocking everything else by default. In testing against eight documented attack primitives, a standard Docker container blocked none, while Ephemora Cell blocked all eight, covering vectors such as shell access, forking, sockets, and filesystem escapes. The project, released under the Apache-2.0 license, supports any language that compiles to WASI/WASM and includes a GitHub Action for sandboxed CI runs. Developers building AI agents or MCP tools can install it via pip and run the included attack scripts to independently verify its security boundaries.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

GPT-6 Astra Independently Builds a Puzzle, Supply-Chain Model, and Booklet Tool

OpenAI's GPT-6 Astra ran three independent AI-agent experiments, published on 4 September, in which separate agents built a string-based puzzle game called Loose End, a supply-chain simulation called The Order Echo, and a single-sheet booklet layout tool called Pocket Press. Each project went through a structured prototype review process before refinement, with a coordinating agent identifying and flagging issues for builders to correct. In Loose End, a logic puzzle involving a finite string, pegs, and bells, an early input bug ignoring brief keypresses was caught and fixed before the game was finalised. The Order Echo models how different ordering policies affect a four-tier supply chain, with its interface allowing users to inspect weekly demand and order arithmetic on a shared scale. Pocket Press arranges an eight-page booklet onto a single A4 or US Letter sheet and, following reviewer feedback, gained discoverable save options, folding instructions, and the ability to reload editable source files.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Thor Commerce Lets Developers Build AI-Assisted Custom Storefronts Without Starting From Scratch

Thor Commerce is a headless commerce platform that separates the commerce backend from the customer-facing storefront, allowing developers to build fully custom shopping experiences without rebuilding core logic. The platform provides two GraphQL APIs — an Admin API for server-side operations and a Storefront API for buyer-facing flows — both generating TypeScript types to keep code aligned with the schema. Developers can use AI coding agents such as Claude Code, Codex, or Cursor to customize a provided Next.js reference storefront, with the agents guided by typed contracts, structured documentation, and an AGENTS.md file. Thor also supports context-aware pricing and availability, making it suitable for businesses running both direct-to-consumer and B2B operations from a single catalog. The tutorial walks developers from local setup through product resolution, cart and checkout flows, AI-assisted customization, and a path to production deployment.

0 comments Read more at DEV Community