[{"data":1,"prerenderedAt":327},["ShallowReactive",2],{"post-en-claude-code-codex-cursor-security-checklist":3,"similar-en-claude-code-codex-cursor-security-checklist":308},{"id":4,"title":5,"authorSlug":6,"body":7,"category":293,"description":13,"extension":294,"image":295,"locale":296,"meta":297,"navigation":298,"path":299,"publishedAt":300,"readingMinutes":301,"seo":302,"siblingSlug":303,"slug":304,"stem":305,"summary":306,"updatedAt":285,"__hash__":307},"blog\u002Fblog\u002Fen\u002Fclaude-code-codex-cursor-security-checklist.md","Claude Code, Codex and Cursor: a security buyer's checklist","sarah-krarup",{"type":8,"value":9,"toc":284},"minimark",[10,14,22,27,30,33,89,112,116,119,135,138,141,145,148,242,245,249,252,255,258,261,265,268,281],[11,12,13],"p",{},"An AI coding agent can read a project, change files and use tools to complete a development task. Choosing one therefore means deciding what access you are comfortable giving it, alongside whether its code is useful.",[11,15,16,17,21],{},"For a security or IT leader, the practical question is: ",[18,19,20],"strong",{},"Can we explain what this agent can read, change and send elsewhere in the setup we intend to use?"," Start there before comparing coding benchmarks.",[23,24,26],"h2",{"id":25},"compare-the-setup-your-team-will-actually-use","Compare the setup your team will actually use",[11,28,29],{},"A product name is too broad for a security decision. A developer working locally, a cloud agent running on a separate machine and an agent connected to company systems can have different access and approval settings.",[11,31,32],{},"Record the product, subscription, interface and integrations being assessed. Then use the vendor's current documentation to check the controls. The following starting points were reviewed on 13 October 2026; they are not a ranking of the products' security.",[34,35,36,52],"table",{},[37,38,39],"thead",{},[40,41,42,46,49],"tr",{},[43,44,45],"th",{},"Product",[43,47,48],{},"Documented starting point",[43,50,51],{},"What to verify in your pilot",[53,54,55,67,78],"tbody",{},[40,56,57,61,64],{},[58,59,60],"td",{},"Claude Code",[58,62,63],{},"Anthropic documents permission modes and a shell sandbox with filesystem and network isolation.",[58,65,66],{},"Which mode is active, which commands can run and which files and network destinations are reachable.",[40,68,69,72,75],{},[58,70,71],{},"Codex",[58,73,74],{},"OpenAI distinguishes sandbox boundaries from the policy governing action approvals.",[58,76,77],{},"The effective boundaries and approval rules for the interface and execution environment you use.",[40,79,80,83,86],{},[58,81,82],{},"Cursor",[58,84,85],{},"Cursor documents agent controls and a Privacy Mode that prevents training on your data when enabled.",[58,87,88],{},"Privacy settings, agent execution controls and the behavior of local and cloud workflows separately.",[11,90,91,92,99,100,105,106,111],{},"Read the details in ",[93,94,98],"a",{"href":95,"rel":96},"https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fsecurity",[97],"nofollow","Claude Code's security guidance",", ",[93,101,104],{"href":102,"rel":103},"https:\u002F\u002Flearn.chatgpt.com\u002Fdocs\u002Fsandboxing",[97],"OpenAI's sandbox documentation"," and ",[93,107,110],{"href":108,"rel":109},"https:\u002F\u002Fcursor.com\u002Fsecurity",[97],"Cursor's security documentation",". Availability and behavior can depend on the plan and configuration, so check the specific setup rather than treating a documentation page as proof that a control is enabled.",[23,113,115],{"id":114},"separate-data-protection-from-permission-to-act","Separate data protection from permission to act",[11,117,118],{},"Two questions often get mixed together during procurement:",[120,121,122,129],"ul",{},[123,124,125,128],"li",{},[18,126,127],{},"Data handling:"," Where does source code, a prompt, an index or an activity log go? What is retained, who can access it and can it be used for training?",[123,130,131,134],{},[18,132,133],{},"Action permissions:"," Can the agent change files, use credentials, contact external services or publish something?",[11,136,137],{},"A favorable answer about training does not establish a safe execution boundary. Conversely, an isolated execution environment does not answer every question about vendor retention or subprocessors.",[11,139,140],{},"Ask the vendor for the applicable data-handling terms and ask your engineering team to demonstrate the effective permissions. Both answers belong in the approval record.",[23,142,144],{"id":143},"use-this-checklist-before-approving-a-coding-agent","Use this checklist before approving a coding agent",[11,146,147],{},"The same questions work across products. A small security team can use the table as a review worksheet, adding an owner and evidence beside each answer.",[34,149,150,163],{},[37,151,152],{},[40,153,154,157,160],{},[43,155,156],{},"Review area",[43,158,159],{},"Question to answer",[43,161,162],{},"Useful evidence",[53,164,165,176,187,198,209,220,231],{},[40,166,167,170,173],{},[58,168,169],{},"Project access",[58,171,172],{},"Which repositories and directories can the agent read or change?",[58,174,175],{},"Repository grants and effective filesystem restrictions.",[40,177,178,181,184],{},[58,179,180],{},"Company data",[58,182,183],{},"Which data classes may employees include in prompts or connected tools?",[58,185,186],{},"Approved data rules and the relevant vendor terms.",[40,188,189,192,195],{},[58,190,191],{},"Network access",[58,193,194],{},"Which destinations can the execution environment contact?",[58,196,197],{},"Network policy and a safe demonstration of a blocked destination.",[40,199,200,203,206],{},[58,201,202],{},"Credentials",[58,204,205],{},"Which identity and credentials are available to the agent?",[58,207,208],{},"Scoped access, credential lifetime and a revocation procedure.",[40,210,211,214,217],{},[58,212,213],{},"Connected tools",[58,215,216],{},"Who approves integrations and their permissions?",[58,218,219],{},"Integration inventory with owners and approved scopes.",[40,221,222,225,228],{},[58,223,224],{},"Consequential actions",[58,226,227],{},"Who approves publishing, destructive changes or access expansion?",[58,229,230],{},"Approval rules and named reviewers.",[40,232,233,236,239],{},[58,234,235],{},"Accountability",[58,237,238],{},"What activity can you investigate afterward?",[58,240,241],{},"Available logs, code-review records and a reporting contact.",[11,243,244],{},"Some controls may sit in your existing development platform rather than the coding agent. Record that dependency. For example, a mandatory human review may be enforced through repository rules; the agent's ability to propose code does not need to include permission to merge it.",[23,246,248],{"id":247},"test-boundaries-with-a-small-realistic-pilot","Test boundaries with a small, realistic pilot",[11,250,251],{},"Choose a representative project with synthetic data and no production credentials. Give the agent an ordinary task, such as updating documentation or proposing a small change with tests.",[11,253,254],{},"Check the proposed changes and observe the permissions it requests. Can it reach an unrelated directory? Does an unapproved network destination stay blocked? Can it propose a pull request while a human retains responsibility for merging it?",[11,256,257],{},"Use harmless test targets and an approved environment. Keep a short record of the configuration, observed behavior and limitations. Repeat the relevant checks when you change execution mode or add an integration.",[11,259,260],{},"Avoid making the pilot a contest for the longest task completed without intervention. You are testing whether the boundaries hold while the tool does useful work.",[23,262,264],{"id":263},"make-the-approval-usable-for-employees","Make the approval usable for employees",[11,266,267],{},"An approval should tell developers which account to use, which projects are in scope, which data is permitted and when they need another person's review. Give them a contact for uncertain requests and an easy way to report unexpected behavior.",[11,269,270,271,275,276,280],{},"Include examples in ",[93,272,274],{"href":273},"\u002Fsolutions\u002Ftraining","security awareness training",": a request for a broader permission, a prompt containing customer data and an unfamiliar instruction found in a repository. Our guide to ",[93,277,279],{"href":278},"\u002Fblog\u002Fprompt-injection-ai-coding-agents","prompt injection in AI coding agents"," provides a short exercise for that conversation.",[11,282,283],{},"Start with the smallest access that makes the pilot useful. Expand it only when an owner can explain the need and verify the new boundary. That gives your team a clear route to adopting the agent without leaving the approval decision to each developer.",{"title":285,"searchDepth":286,"depth":286,"links":287},"",2,[288,289,290,291,292],{"id":25,"depth":286,"text":26},{"id":114,"depth":286,"text":115},{"id":143,"depth":286,"text":144},{"id":247,"depth":286,"text":248},{"id":263,"depth":286,"text":264},"Tips","md","\u002Fimages\u002Fblog\u002Fclaude-code-codex-cursor-security-checklist-pexels-5908723.jpg","en",{},true,"\u002Fblog\u002Fen\u002Fclaude-code-codex-cursor-security-checklist","2026-10-13",5,{"title":5,"description":13},"claude-code-codex-cursor-sikkerhedstjekliste","claude-code-codex-cursor-security-checklist","blog\u002Fen\u002Fclaude-code-codex-cursor-security-checklist","Compare AI coding agents by their access to code, data and tools. Use a practical checklist to approve the right setup for your team.","x_uIv3MzZOu1V26D8t-r1MpaSejO6RmTCuZuSattQa8",[309,315,320],{"slug":310,"title":311,"summary":312,"category":293,"publishedAt":300,"image":313,"readingMinutes":314},"shadow-ai-audit","Shadow AI audit: a 14-day plan for growing companies","Find unapproved AI use, assess the real risks and give employees safer options with a practical 14-day shadow AI audit for a growing company.","\u002Fimages\u002Fblog\u002Fshadow-ai-audit-pexels-8127811.jpg",6,{"slug":316,"title":317,"summary":318,"category":293,"publishedAt":300,"image":319,"readingMinutes":301},"vibe-coding-at-work-ai-use-policy","Vibe coding at work: an AI use policy your team can follow","Give employees clear rules for building apps with AI. Adapt a practical policy covering approved tools, company data, access and review before release.","\u002Fimages\u002Fblog\u002Fvibe-coding-at-work-ai-use-policy-pexels-7670739.jpg",{"slug":321,"title":322,"summary":323,"category":293,"publishedAt":324,"image":325,"readingMinutes":326},"what-is-micro-learning","What is micro-learning?","Learn more about what micro-learning is, how it can be used in different contexts, and the benefits and challenges associated with this learning method.","2024-09-18","https:\u002F\u002Fedge-cdn.moxso.com\u002Fblog\u002Fb6H2FyhhXRkVChtMjGjr9zKRQHA6JPMeqBHIyRqY.jpg",9,1792499638779]