SShortSingh.
Back to feed

Firecrawl Local: Single Rust Binary Offers Infrastructure-Free Alternative to Self-Hosted Stack

0
·24 views

Firecrawl has released Firecrawl Local, a single Rust binary that scrapes, crawls, and maps websites to Markdown without requiring a database, queue, or containers. It serves the same v2 API on localhost port 3002 as the traditional self-hosted Docker Compose stack. The lighter option omits features needing heavier infrastructure, such as LLM-based json and summary formats, persistent job storage, and screenshots, and it routes search to Firecrawl Cloud. Both variants are open source under AGPL-3.0 and live in the same GitHub repository, with Local complementing rather than replacing the full server.

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 ·

glab mr view now reports 'unknown' for merge request conflicts while GitLab is still checking

A change to the GitLab CLI (glab) addresses a gap where redirecting mr view output to a file dropped the conflict status shown in the interactive view. The plain-text output now prints a conflicts line with true, false, or unknown, where unknown covers the pending merge-status states unchecked, checking, and preparing. The formatter checks detailed_merge_status before rendering the boolean, since a false value during calculation could mislead scripts. Automation is advised to read the existing JSON fields instead of grepping the text output, and to retry within a bounded timeout when the status is pending. Regression tests cover the true, false, and pending cases, and the change was verified with passing tests and GitLab CI.

0
ProgrammingDEV Community ·

Scanner Dependency Alerts in Code Review: Blind Judges Find No Fully Real Issues

A developer tested whether vulnerability findings from dependency scanners could go directly into an AI code review report without a model verifying them. The reviewer, a Claude Code plugin for reviewing whole backends, runs per-language static tools such as govulncheck, cve-lite and osv-scanner in a Docker image. In the early setup, a model read every advisory and a verifier checked the findings, which cost about 21% of total run cost and sometimes dropped scanner-reported advisories. The replacement used fixed rules that tiered findings by whether the vulnerable code was called or imported, skipping the verifier. Blind Opus judges then reviewed the findings across four private projects, and of 10 distinct issues none was fully real, falling to 0 of 5 after a fix that halved the noise.

0
ProgrammingDEV Community ·

Average time for attackers to exploit flaws after patches has collapsed to near zero

Google's Mandiant team has tracked how long attackers take to exploit a vulnerability after its patch is released, and the figure fell from 63 days in 2018-2019 to 5 days in 2023. The 2024 figure was negative, meaning attacks began before patches existed, and a further drop is estimated for 2025. The article ties this trend to advances in AI, citing a DARPA AI Cyber Challenge final in August 2025, where automated systems found planted and unplanted bugs at about $152 per task. It also cites Anthropic's April 2026 announcement that its model had found thousands of zero-day flaws, with more than 10,000 high or critical findings by late May. The author argues that human-paced patching cannot keep up with automated discovery and proposes defenses that block any behavior not previously observed.

0
ProgrammingDEV Community ·

JunctionBox AI runs Claude Code, Codex, Grok and Ollama agents side by side on your own machine

JunctionBox AI is a freemium desktop app for Linux, macOS and Windows that lets users run several coding agents together without needing an account. Agents can use existing Claude Code, Codex or Grok CLI sign-ins, API keys for Anthropic, OpenAI and xAI, or local and remote Ollama and llama.cpp servers. Users can compare answers from multiple agents to the same prompt and group agents into teams that pass work between them. Scheduled, webhook or Telegram-triggered jobs, run history, and a sandbox with permission profiles are available on Linux and macOS, while Windows agents currently run with the user's full account rights. The app also includes a TV launcher for Roku, Google TV and Fire TV that lets users control agents and remote machines from the couch behind a PIN.