More reliable AI coding
Remembers, plans first, checks before edits
- AI backend code that guards against races, leaks and bad migrationsCodingbackendskillsClaude CodePostgreSQLFree; about 2Γ tokens per taskβ Partly testedOverseas
- An afternoon of AI work, a 3D action game you play in the browserCodingClaude CodeTripoJimengSkill free; Claude quota plus asset toolsβ Partly testedChina
- Keep a long-running AI-built project from turning into a messCodingClaudeCodexoil-delegateNeeds Claude and Codex plansβ Partly testedChina
- Scan a skill before installing, then read the flagged linesCodingSkillSpectoruvFree; static scan needs no modelβ Tested by usChina
- Have fresh-context AI reviewers poke holes before you codeCodingEGMCodexClaude CodePlugin free; reviews use your model quotaβ Partly testedOverseas
- 1,050 memory recalls audited: what makes agent memory pay offCodingNowledge MemCodexClaude CodeFree tier: 50 memories; review uses model quotaβ Not tested by usChina
- A designer who doesn't code ships an iOS app with six agentsCodingClaude CodeXcodeascFree plugin; needs Claude plan and Apple developer accountβ Partly testedOverseas
- Locate a mystery beep with phones and tabletsCodingClaude CodeBeep LocatorPythonFree; needs Python and several devicesβ Partly testedOverseas
- Security audits with isolated hunter and verifier agentsCodingCloudflaresecurity-auditClaude CodeSkill free; full audits spend model quotaβ Partly testedOverseas
- One Git repo syncs the team's skills and MCPCodingTeamAIClaude CodeGitFree; needs Git and an agent you already useβ Partly testedChina
- Static skill inspection that blocks claim mismatchesCodingUSB+secscanMCPFree; secscan never executes the skillβ Partly testedOverseas
- Read the code, then fix calqued jargonCodingclear-namingCodexClaude CodeSkill free; no extra runtimeβ Partly testedChina
- Align terms, check facts, then decideCodingsinfire-skillsCodexgrill-meSkills free; needs Codex or Claudeβ Partly testedChina
- Skip subagents for ordinary code search: up to +221% cost, no recall gainCodingClaude CodeTokenSaveSubagentsStudy open; running the agent still needs a planβ Partly testedOverseas
- Click the desktop via the a11y tree, not pixelsCodingagent-desktopComputer UseRustCLI free; needs local a11y permissionβ Partly testedOverseas
- One command to see if your agent overstepsCodingiFixAiPythonClaude CodeOpen source; real audits pay for a judge modelβ Partly testedOverseas
- Dozens of AI file edits β one business-readable report pageCodingoil-diffClaude CodeGitFree MIT; needs Python 3.10+ and gitβ Partly testedChina
- Make scraping agents give up on turn 1CodingClaude CodeCloudflare WorkersSEOSkill free; needs your agent planβ Partly testedOverseas
- See every wasted lookup Claude Code makesCodingClaude CodeModsNeeds Claude plan; Mod API earlyβ Not tested by usOverseas
- Four questions that make AI go deeper, not biggerProductivityClaude CodeCodexFree β just a promptβ Not tested by usOverseas
- One lead agent dispatching the restCodinglbkitHerdrchattaFree tools; several AI plansβ Not tested by usChina
- Claude Code that remembers after compaction and proves "done"CodingClaude CodeFree scripts; Claude planβ Partly testedOverseasEditor's pick
- Keep passwords out of the chat when driving AI by phoneCodingCodexTailscaleFree tools; agent plan extraβ Partly testedOverseas
- Turn your own agent chats into tests, find a half-price setupCodingClaude CodeCodexHarborFree tools; pay for model callsβ Partly testedOverseas
- Three prompts to a LaTeX tool that lives in the linkCodingClaude CodeNeeds Claude Code; free hostingβ Partly testedOverseas
- Ship an app without coding by splitting work across AIsCodingClaude CodeOpenCodeDeepSeekFree template; model plans extraβ Partly testedChina
- Block agent rule-breaking before each commitCodingGitFree; built into gitβ Partly testedOverseas
- Export a whole Claude Code chat for the next agentCodingClaude CodeFree command; Claude planβ Partly testedOverseas
- One prompt that makes coding agents think before writingCodingClaude CodeCodexFree; a little more usageβ Not tested by usOverseas
- Let the AI question you before it codesCodingClaude CodeCodexFree; a few minutes of Q&Aβ Not tested by usOverseas
- Make the coding agent check before it editsCodingCursorpstackFree plugin; extra usageβ Not tested by usOverseas
Each one lists the steps, cost and original author; open it for the full how-to. "Tested by us / Partly tested / Not tested by us" says whether we tried it ourselves.
Follow for updates
Only AI use cases someone actually tried. Updated when there is something worth adding, not on a schedule.