By job / Coding

Ship software faster

26 tools for coding, in the order I'd try them.

1. 1.

   Cursor in coding

   The editor most builders end up in: codebase-aware edits that actually land.

   Free tier; Pro from $20/mo

2. 2.

   Claude in coding

   My pick for anything long: contracts, big codebases, a messy brief. It reads carefully and says when it isn't sure.

   Free tier; Pro from $20/mo

3. 3.

   GitHub Copilot in coding

   Safe default for teams already on GitHub; strong autocomplete, solid chat.

   Free tier; Pro $10/mo

4. 4.

   Devin Desktop in coding

   Was Windsurf; Cognition rebranded it as Devin Desktop. Same IDE, now the command center for local and cloud coding agents. Update in-app and your settings stay put.

   Free tier; Pro from about $20/mo (legacy Windsurf plans carry over)

5. 5.

   Replit in coding

   Idea to a running URL without installing anything. The code gets messy if you keep going, so treat it as a sketchpad.

   Free tier; paid plans available

6. 6.

   Lovable in coding

   Type what you want, get a working app. Good enough to show customers before you hire anyone.

   Free tier; paid from $25/mo

7. 7.

   v0 in coding

   Best for generating clean React UI you can actually keep.

   Free tier; paid plans available

8. 8.

   Bolt.new in coding

   Similar to Lovable, a little rougher, sometimes quicker. Try both on the same prompt and keep the one you like.

   Free tier; paid plans available

9. 9.

   OpenCode in coding

   Open-source coding agent for terminal, IDE and desktop with multi-session support and 75+ model providers. A practical default when you want an agent without locking into one lab’s CLI.

   Open source free; optional Zen paid model access; BYO provider keys

10. 10.

    Replicas in coding

    Runs Claude Code, Codex, Cursor or OpenCode in their own cloud VMs and hands back a PR, a reply or a screen recording. It's for teams already juggling several local agents who want them triggered from Slack, Linear or GitHub instead.

    $100 free usage (no card); Developer $50/mo incl. $50 usage, Team from $200/mo; per-minute workspace billing

11. 11.

    LightSprint in coding

    Lets PMs, designers and ops staff describe a change, agree a plan, preview it and open a PR on the real codebase, while engineers still approve every merge. Most useful where the backlog is full of small changes waiting on one or two developers.

    Free ($0, 5 tasks/plans a week, BYO model key); Pro $40/seat/month; Enterprise custom

12. 12.

    Polylane in coding

    Watches your logs and infra, investigates production drift, and opens a fix PR before anyone gets paged. Impact Analysis also reviews pull requests against live capacity and telemetry — useful when coding agents ship faster than your on-call rotation can keep up.

    Free plan; paid self-serve from about $80/mo (see polylane.com/pricing)

13. 13.

    Klarent in coding

    Autonomous QA agents that use LLMs to author deterministic tests for web and mobile, so AI-written code gets repeatable coverage. Fits eng and ecommerce teams drowning in brittle automation.

    Outcome-oriented QA platform (contact sales; human-in-the-loop verification)

14. 14.

    Kerno in coding

    Runtime QA engineer for backends: maps your stack, tests critical flows and MCP tools against the real database and queues, then re-checks every PR. Catches the auth, contract and security bugs unit tests miss when agents write the code.

    Self-serve plans (contact for enterprise); free trial / get-started via site

15. 15.

    Reason in coding

    Autonomous coding agent that connects to GitHub, takes an issue, and opens evidence-backed pull requests with logs and screen recordings. Positions itself as cheaper per solved task than Codex while still usable from Mac, web, terminal or Slack.

    Free to start locally or in the cloud; paid usage for cloud agents (YC credits available) · limited in Asia

16. 16.

    Agnost AI in coding

    Reads production conversations to spot when users correct your agent, repeat themselves or give up, groups the failures, and suggests a fix it can test before you ship. It's useful once a support or sales agent is live and your dashboards say everything is fine.

    Free (1,000 events/month); Starter $49/month; Pro $499/month; Enterprise custom

17. 17.

    Offrun in coding

    Mac workspace that runs several coding agents in parallel, each in its own git worktree, with usage meters and one-click switch when a login hits its limit. Peer review uses one of your own agent accounts; Offrun never sees your keys.

    Free and unlimited; bring your own Claude Code, Codex, AGY or Grok Build subscriptions

18. 18.

    Inferock in coding

    Puts an independent receipt on every LLM API call so you can see what you were billed versus what failed or retried. Serve adds serverless inference with money back when calls fail — handy when model spend is the line item you argue about.

    Watch for receipts on your own keys; Serve is prepaid pay-per-token with credits back on failed calls

19. 19.

    Rill Browser in coding

    A calm Mac browser that sends whatever you're looking at to Claude Code or Codex with ⌘E, groups tabs and turns history into searchable days. Free, no API keys, so it only makes sense if you already pay for one of those agents.

    Free (uses your existing Claude Code or Codex plan); macOS 14+, Apple silicon

20. 20.

    Agent37 in coding

    Cheap persistent sandboxes for Hermes, OpenClaw, Claude Code and similar agents. Compute sleeps with the box; model tokens are separate. Best if you already run agents and just need a host that does not dominate the bill.

    $5 starter credit after card; then ~$0.80/vCPU + $0.70/GB RAM + $0.09/GB disk per 730-hour month, metered by the minute (typical small sandbox ~$0.37–$4.76/mo depending on hours)

21. 21.

    FastRouter.ai in coding

    OpenAI-compatible gateway that routes each request to the right model for cost, latency and quality, with failover and observability. Saves small teams from hard-coding one expensive provider.

    Free to start; pay for routed usage across 200+ models with 0% markup on tokens (see fastrouter.ai for tiers)

22. 22.

    CodeCrab in coding

    Local-first desktop PR reviewer: code never leaves the laptop, reviews map onto the diff, and fixes stay on your branch after your own tests run. Built for teams that refuse cloud review bots on private repos.

    Free public beta; uses your existing Claude Code / Cursor / Copilot CLI subscriptions

23. 23.

    Tabnine in coding

    Pick it when privacy and self-hosting matter more than raw power.

    Paid plans; see site

24. 24.

    Tiden in coding

    Keeps product specs beside the code and will not close an agent session while a promise lacks a passing test. Fits teams that already ship with Claude Code, Cursor, Copilot or Codex and need intent, not just green CI.

    Install via curl; cloud signup at app.tiden.ai (check site for current plan tiers)

25. 25.

    Context.dev in coding

    One API for scraping pages to clean Markdown, crawling sites, web and news search, and pulling a company's logo, colours and fonts from its domain. JS rendering and proxies are included, which saves a lot of setup for small teams building agents.

    Free 1,000 credits/mo; Developer $19/mo; Pro $99/mo; Growth $299/mo; Scale $499/mo

26. 26.

    Whacka in coding

    No-code AI app builder: describe an app in plain words and it builds, hosts and publishes it with logins, database, payments, push notifications and built-in AI. Templates cover bookings, shops, POS, inventory and classes. Pro plans ship to Google Play and the App Store.

    Free (24 credits/month, apps online 14 days); Weekly $3/week; Starter $8/month, Pro $32/month (native iOS/Android, own domain, code export), Ultra $96/month, all billed annually; credit top-ups available

The top spot here can be sponsored. Details.