[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AX9GZQF8YAPYAJENWTR561":3,"skill-same:01M7AX9GZQF8YAPYAJENWTR561":64},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":30,"ownerHtmlUrl":31,"ownerBlogUrl":30,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7AX9GZQF8YAPYAJENWTR561","security-audit",null,"Offensive security audit of UZE - a red team of agents, each in a bubblewrap sandbox with the operator's home hidden, attacking one surface (acquisition, delivery, extension scripts, workspace and agents, terminal runtime, parsers, distribution) with working exploits, then every finding recorded and fixed as an `audit-\u003Csubject>` OpenSpec change that ends in proof. Use when someone asks for a security audit, a pentest, a red team, threat modelling, \"procura vulnerabilidade\", \"tenta quebrar o uze\", before a release that widens what UZE runs or trusts, after a change to trust\u002Fconsent\u002Fconfinement\u002Fapproval, or to re-verify an earlier audit's fixes. Not for a general code review (that is `code:polish`) or one dependency's provenance (the `dependencies` rule).","uze-sh\u002Fuze","uze-sh","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F334211959?v=4","01M7AX5KNT81AC56YMRT586MBW","uze","01M7AX5KPDQA2QP7DWGZFMH4PQ",4,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fsecurity-audit\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: security-audit\ndescription: Offensive security audit of UZE - a red team of agents, each in a bubblewrap sandbox with the operator's home hidden, attacking one surface (acquisition, delivery, extension scripts, workspace and agents, terminal runtime, parsers, distribution) with working exploits, then every finding recorded and fixed as an `audit-\u003Csubject>` OpenSpec change that ends in proof. Use when someone asks for a security audit, a pentest, a red team, threat modelling, \"procura vulnerabilidade\", \"tenta quebrar o uze\", before a release that widens what UZE runs or trusts, after a change to trust\u002Fconsent\u002Fconfinement\u002Fapproval, or to re-verify an earlier audit's fixes. Not for a general code review (that is `code:polish`) or one dependency's provenance (the `dependencies` rule).\ncompatibility: Linux with bubblewrap (bwrap) for the lab; the OpenSpec CLI for the audit schema.\n---\n\n# Security audit\n\nUZE runs on the operator's machine, beside the agents it launches, with the\noperator's permissions. Every boundary it draws - plugin trust, extension\nconsent, project-command approval, the pane\u002Foperator split, the Rhai engine's\nbounds - is a claim an attacker will test. This skill tests them the way an\nattacker would: with working exploits against the real binary, in a sandbox\nthat cannot reach the operator's real home, and it carries what breaks to a\nfix that is proven by re-running the same exploit.\n\nIt has two halves. **Attack** ends in a findings report. **Fix** turns each\ngroup of findings into an `audit-\u003Csubject>` OpenSpec change on the `audit`\nschema (`findings → proposal → specs → design → tasks → evidence`) and works\nit until `evidence.md` closes every finding.\n\n## Step 1 - scope and baseline\n\n- Record the commit under audit (`git rev-parse HEAD`) and the date.\n- Read `references\u002Fsurface.md`: the attack surfaces, the threat\n  actors, what earlier audits found and the defenses that held. A defense\n  that held is re-tested only if its code changed since\n  (`git log \u003Clast-audit-commit>.. -- \u003Cpaths>`).\n- Read `docs\u002Farchitecture\u002Finvariants.md` and the ADRs of each surface: the\n  claims to break.\n- Archived `openspec\u002Fchanges\u002Farchive\u002F*-audit-*` are earlier audits; their\n  `evidence.md` names the regression tests that must still pass.\n\n## Step 2 - the lab\n\nNever run uze or an exploit against the operator's home: it holds the real\n`~\u002F.uze`, harness configs and keys.\n\n```bash\nAUDIT=\u003Cscratchpad>\u002Faudit; mkdir -p \"$AUDIT\u002Fbin\"\ncargo build --locked --bin uze && cp target\u002Fdebug\u002Fuze \"$AUDIT\u002Fbin\u002Fuze\"\ncp .agents\u002Fskills\u002Fsecurity-audit\u002Fscripts\u002Fsandbox \"$AUDIT\u002Fbin\u002Fsandbox\"\ncp .agents\u002Fskills\u002Fsecurity-audit\u002Freferences\u002Fbrief.md \"$AUDIT\u002FBRIEF.md\"   # fill in AUDIT\nSRC=$PWD \"$AUDIT\u002Fbin\u002Fsandbox\" smoke -- bash -c 'ls -A \u002Fhome\u002F* \u002Fmnt \u002Frun; uze --version'\n# your home shows only .cargo\u002F.rustup; \u002Fmnt and \u002Frun are empty\n```\n\nThe smoke test must show the real home, the other mounts and the\nprivileged sockets hidden. One box per auditor\n(`$AUDIT\u002Fboxes\u002F\u003Cbox>`), so PoCs never interfere.\n\n## Step 3 - the red team\n\nOne agent per surface in `references\u002Fsurface.md`, in parallel (keep it under ten),\neach briefed with `$AUDIT\u002FBRIEF.md` plus the surface's own paragraph: its\ncode, its ADRs, and the hypotheses to try. Auditors only read the repo and\nwrite in their box. They must not run `uze agent work name` or any git\nwrite: a subagent inherits the parent's agent stamp and would rename the\nparent's branch. Say so in the brief.\n\nA subagent cannot write files outside its box and its report can arrive\ntruncated. Ask for the report inline in the final message, in the brief's\nfixed shape. If a report is cut off, ask the agent to resend the rest.\n\n## Step 4 - verify before you believe\n\nAn auditor's report is a claim. Before it reaches the operator:\n\n- Re-run every High\u002FCritical PoC yourself in a fresh box\n  (`$AUDIT\u002Fboxes\u002Fverify-\u003Cid>`), copying only the PoC's inputs.\n- For a PLAUSIBLE finding, read the exact code path it names.\n- Drop what you cannot reproduce or locate. Say which you verified\n  yourself.\n\nReport to the operator grouped by root cause, not by auditor. A table per\ntheme, with the structural fix next to the point fixes.\n\n## Step 5 - record: in private first, in the repository only with the fix\n\nThis repository is public, and a committed finding is a disclosure that\nsurvives deletion. The `security-findings` rule (`.agents\u002Frules\u002F`) holds the\npolicy. In short:\n\n- **Every verified finding opens as a private draft GitHub Security\n  Advisory.** Make one per root cause, with the finding in the\n  `findings.md` shape. The rule gives the `gh api` call. Check that its\n  state reads `draft`.\n- **A finding enters the repository only on the branch that fixes it.**\n  It goes in as `openspec\u002Fchanges\u002Faudit-\u003Csubject>\u002Ffindings.md`, in the same\n  delivery as the fix.\n\nGroup the findings by the property that does not hold, not by crate. For\nexample, \"consent covers what is declared, not what runs\". Each group is\none advisory and, when fixed, one change:\n\n```bash\nopenspec new change audit-\u003Csubject> --schema audit\n```\n\nWrite `findings.md` from the advisory text. It ends with `## Defenses that\nheld`. Then write proposal, specs, design and tasks as for any change. Make\nthe regression test and the exploit re-run explicit tasks.\n\nSpecs state the property, never the attack. That makes them safe to commit\nbefore anything else.\n\n## Step 6 - fix, in parallel checkouts\n\nOne agent per change, each in its own checkout from\n`uze agent work split \u003Ctopic>`. Brief each with:\n\n- the change path;\n- the `openspec-apply-change` contract;\n- the rule that every finding gets a test failing on the audited commit;\n- the lab path, so it can rebuild the binary and re-run the original PoC.\n\nJoin each with `uze agent work join \u003Ctopic>` when it is green.\n\n## Step 7 - evidence, then archive\n\n`evidence.md` closes each finding with four things:\n\n- the behavior change and where it lives;\n- the regression test;\n- the exploit re-run against the fixed binary, with its output;\n- the residual.\n\nA residual is either an **accepted limit**, which the threat model excludes\nby design and is stated without a recipe, or **unfixed**. An unfixed\nresidual, such as a bypass found while fixing, goes to a new private\nadvisory, and `evidence.md` names only its id. The finding it belongs to is\nthen partially closed, and says so.\n\nThen run the gate:\n\n- `cargo test --workspace`\n- `cargo clippy --all-targets -- -D warnings`\n- `openspec validate --all --strict`\n\nAn `audit-*` change is archived only when evidence closes every finding\n(`openspec\u002Fconfig.yaml`, archive guidance). Every `audit-*` change has a\ndraft advisory, opened when the fix is ready (the `security-findings`\nrule). Publish it right after the release that carries the fix.\n\nFinally, update `references\u002Fsurface.md`. Add the new changes and advisory ids (ids only) to its index and new\ndefenses to \"held\". If an auditor tested a surface that is missing from it,\nadd that surface.\n","# Security audit\n\nUZE runs on the operator's machine, beside the agents it launches, with the\noperator's permissions. Every boundary it draws - plugin trust, extension\nconsent, project-command approval, the pane\u002Foperator split, the Rhai engine's\nbounds - is a claim an attacker will test. This skill tests them the way an\nattacker would: with working exploits against the real binary, in a sandbox\nthat cannot reach the operator's real home, and it carries what breaks to a\nfix that is proven by re-running the same exploit.\n\nIt has two halves. **Attack** ends in a findings report. **Fix** turns each\ngroup of findings into an `audit-\u003Csubject>` OpenSpec change on the `audit`\nschema (`findings → proposal → specs → design → tasks → evidence`) and works\nit until `evidence.md` closes every finding.\n\n## Step 1 - scope and baseline\n\n- Record the commit under audit (`git rev-parse HEAD`) and the date.\n- Read `references\u002Fsurface.md`: the attack surfaces, the threat\n  actors, what earlier audits found and the defenses that held. A defense\n  that held is re-tested only if its code changed since\n  (`git log \u003Clast-audit-commit>.. -- \u003Cpaths>`).\n- Read `docs\u002Farchitecture\u002Finvariants.md` and the ADRs of each surface: the\n  claims to break.\n- Archived `openspec\u002Fchanges\u002Farchive\u002F*-audit-*` are earlier audits; their\n  `evidence.md` names the regression tests that must still pass.\n\n## Step 2 - the lab\n\nNever run uze or an exploit against the operator's home: it holds the real\n`~\u002F.uze`, harness configs and keys.\n\n```bash\nAUDIT=\u003Cscratchpad>\u002Faudit; mkdir -p \"$AUDIT\u002Fbin\"\ncargo build --locked --bin uze && cp target\u002Fdebug\u002Fuze \"$AUDIT\u002Fbin\u002Fuze\"\ncp .agents\u002Fskills\u002Fsecurity-audit\u002Fscripts\u002Fsandbox \"$AUDIT\u002Fbin\u002Fsandbox\"\ncp .agents\u002Fskills\u002Fsecurity-audit\u002Freferences\u002Fbrief.md \"$AUDIT\u002FBRIEF.md\"   # fill in AUDIT\nSRC=$PWD \"$AUDIT\u002Fbin\u002Fsandbox\" smoke -- bash -c 'ls -A \u002Fhome\u002F* \u002Fmnt \u002Frun; uze --version'\n# your home shows only .cargo\u002F.rustup; \u002Fmnt and \u002Frun are empty\n```\n\nThe smoke test must show the real home, the other mounts and the\nprivileged sockets hidden. One box per auditor\n(`$AUDIT\u002Fboxes\u002F\u003Cbox>`), so PoCs never interfere.\n\n## Step 3 - the red team\n\nOne agent per surface in `references\u002Fsurface.md`, in parallel (keep it under ten),\neach briefed with `$AUDIT\u002FBRIEF.md` plus the surface's own paragraph: its\ncode, its ADRs, and the hypotheses to try. Auditors only read the repo and\nwrite in their box. They must not run `uze agent work name` or any git\nwrite: a subagent inherits the parent's agent stamp and would rename the\nparent's branch. Say so in the brief.\n\nA subagent cannot write files outside its box and its report can arrive\ntruncated. Ask for the report inline in the final message, in the brief's\nfixed shape. If a report is cut off, ask the agent to resend the rest.\n\n## Step 4 - verify before you believe\n\nAn auditor's report is a claim. Before it reaches the operator:\n\n- Re-run every High\u002FCritical PoC yourself in a fresh box\n  (`$AUDIT\u002Fboxes\u002Fverify-\u003Cid>`), copying only the PoC's inputs.\n- For a PLAUSIBLE finding, read the exact code path it names.\n- Drop what you cannot reproduce or locate. Say which you verified\n  yourself.\n\nReport to the operator grouped by root cause, not by auditor. A table per\ntheme, with the structural fix next to the point fixes.\n\n## Step 5 - record: in private first, in the repository only with the fix\n\nThis repository is public, and a committed finding is a disclosure that\nsurvives deletion. The `security-findings` rule (`.agents\u002Frules\u002F`) holds the\npolicy. In short:\n\n- **Every verified finding opens as a private draft GitHub Security\n  Advisory.** Make one per root cause, with the finding in the\n  `findings.md` shape. The rule gives the `gh api` call. Check that its\n  state reads `draft`.\n- **A finding enters the repository only on the branch that fixes it.**\n  It goes in as `openspec\u002Fchanges\u002Faudit-\u003Csubject>\u002Ffindings.md`, in the same\n  delivery as the fix.\n\nGroup the findings by the property that does not hold, not by crate. For\nexample, \"consent covers what is declared, not what runs\". Each group is\none advisory and, when fixed, one change:\n\n```bash\nopenspec new change audit-\u003Csubject> --schema audit\n```\n\nWrite `findings.md` from the advisory text. It ends with `## Defenses that\nheld`. Then write proposal, specs, design and tasks as for any change. Make\nthe regression test and the exploit re-run explicit tasks.\n\nSpecs state the property, never the attack. That makes them safe to commit\nbefore anything else.\n\n## Step 6 - fix, in parallel checkouts\n\nOne agent per change, each in its own checkout from\n`uze agent work split \u003Ctopic>`. Brief each with:\n\n- the change path;\n- the `openspec-apply-change` contract;\n- the rule that every finding gets a test failing on the audited commit;\n- the lab path, so it can rebuild the binary and re-run the original PoC.\n\nJoin each with `uze agent work join \u003Ctopic>` when it is green.\n\n## Step 7 - evidence, then archive\n\n`evidence.md` closes each finding with four things:\n\n- the behavior change and where it lives;\n- the regression test;\n- the exploit re-run against the fixed binary, with its output;\n- the residual.\n\nA residual is either an **accepted limit**, which the threat model excludes\nby design and is stated without a recipe, or **unfixed**. An unfixed\nresidual, such as a bypass found while fixing, goes to a new private\nadvisory, and `evidence.md` names only its id. The finding it belongs to is\nthen partially closed, and says so.\n\nThen run the gate:\n\n- `cargo test --workspace`\n- `cargo clippy --all-targets -- -D warnings`\n- `openspec validate --all --strict`\n\nAn `audit-*` change is archived only when evidence closes every finding\n(`openspec\u002Fconfig.yaml`, archive guidance). Every `audit-*` change has a\ndraft advisory, opened when the fix is ready (the `security-findings`\nrule). Publish it right after the release that carries the fix.\n\nFinally, update `references\u002Fsurface.md`. Add the new changes and advisory ids (ids only) to its index and new\ndefenses to \"held\". If an auditor tested a surface that is missing from it,\nadd that surface.\n",{"name":5,"description":7,"compatibility":24},"Linux with bubblewrap (bwrap) for the lab; the OpenSpec CLI for the audit schema.","main","e2cfbb55eec2aa007135f9a2256b73f7170a3b9e",".agents\u002Fskills\u002Fsecurity-audit",1,"https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze","https:\u002F\u002Fuze.sh","https:\u002F\u002Fgithub.com\u002Fuze-sh","🧩 Agents come and go. Your work stays.","2026-11-13 05:15:37",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add uze-sh\u002Fuze --skill security-audit",{"tool":40,"command":41},"bunx","bunx skills add uze-sh\u002Fuze --skill security-audit",{"tool":43,"command":44},"pnpm","pnpm dlx skills add uze-sh\u002Fuze --skill security-audit","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fsecurity-audit\u002FSKILL.md",true,[48,54,58],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"references\u002Fbrief.md","REFERENCE","参考文档",5445,"https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fsecurity-audit\u002Freferences\u002Fbrief.md",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002Fsurface.md",7092,"https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fsecurity-audit\u002Freferences\u002Fsurface.md",{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},"scripts\u002Fsandbox","SCRIPT","脚本",2482,"https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fsecurity-audit\u002Fscripts\u002Fsandbox",[65,72,79,86,93,100],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7AXDFA81RA77NHGYVZRCH0E","commit","North Star fixture skill — used to prove native\u002Fgenerated delivery across harnesses.","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fgolden\u002Fmarketplace\u002Fplugins\u002Fflow\u002Fskills\u002Fcommit\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7AXCYQ5WMH46JRJVF57FTHA","oc-skill","opencode native","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fforeign\u002Fopencode\u002Fnative-plugin\u002Fskills\u002Foc-skill\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7AXCVSFQ76DR4GXGJ1Q9PBA","uze-plugin-first","Emits the deterministic UZE Plugin First conformance proof.","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fforeign\u002Fcodex\u002Fnative-plugin\u002Fskills\u002Fuze-plugin-first\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7AXCTD57QJ3GEN54W5PREV2","hello","says hello","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fforeign\u002Fclaude\u002Fnative-plugin\u002Fskills\u002Fhello\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":19},"01M7AXCR1HKRHJ7PP39JQS8SQ5","agy-native","antigravity native","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fforeign\u002Fantigravity\u002Fnative-plugin\u002Fskills\u002Fagy-native\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":19},"01M7AXCNTR4WXE4VT89MK4XQST","review","Review code for correctness and missing tests","https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Ftests\u002F_fixtures\u002Fcanonical\u002Fworkflow\u002Fskills\u002Freview\u002FSKILL.md",[],[]]