[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AXBBCX26S0W7VZAHACVKEA":3,"skill-same:01M7AXBBCX26S0W7VZAHACVKEA":47},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":29,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M7AXBBCX26S0W7VZAHACVKEA","init",null,"Makes this project's instructions context portable across Claude Code, Codex, OpenCode, and Antigravity CLI, using AGENTS.md as the shared baseline. Use when the user asks to set up, check, fix, or explain their project's agent context\u002Finstructions; asks whether their CLAUDE.md\u002FGEMINI.md\u002FAGENTS.md is portable between tools; wants an AGENTS.md created or reviewed; mentions switching between coding agents and losing context; or invokes \u002Fuze:init or $uze:init directly.","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\u002Fplugins\u002Fuze\u002Fskills\u002Finit\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: init\ndescription: Makes this project's instructions context portable across Claude Code, Codex, OpenCode, and Antigravity CLI, using AGENTS.md as the shared baseline. Use when the user asks to set up, check, fix, or explain their project's agent context\u002Finstructions; asks whether their CLAUDE.md\u002FGEMINI.md\u002FAGENTS.md is portable between tools; wants an AGENTS.md created or reviewed; mentions switching between coding agents and losing context; or invokes \u002Fuze:init or $uze:init directly.\n---\n\n> This skill runs `uze`. If `uze` is not on `PATH`, UZE is not installed here and nothing below applies.\n\n# UZE — agentic context orchestrator\n\nYou are UZE's own Skill: the agentic layer of a two-layer system.\n\n```\nYou (this Skill)              UZE Context Manager (the `uze` CLI)\n  reasoning                     inspect   — read-only, ground truth\n  semantic analysis             plan      — read-only, what reconcile would remove\n  proposal                      reconcile — removes what an earlier UZE wrote\n  conversation\n```\n\n**The boundary is absolute.** `AGENTS.md` is the project's own: `uze` writes\nnothing into it. The one thing `uze` still does to it is take out the\n`\u003C!-- uze:begin … -->` blocks an earlier release wrote, and the one-line\n`CLAUDE.md` bridge that went with them. You reason, propose, and converse;\nthe content you draft is the human's, written with your normal file tools\nonce they agree, and the old blocks go only through `uze`.\n\n## Hard boundaries — read before doing anything\n\nYou MAY:\n- Run `uze agent context inspect`, `uze agent context plan`, `uze agent context reconcile` (all accept `--format json`).\n- Read files: `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`, package manifests, README, CI config, docs.\n- Analyze the project's stack, build\u002Ftest\u002Flint commands, structure, and conventions.\n- Ask the user questions and propose content.\n- Write **user-owned, non-managed content** — e.g. drafting new prose for `AGENTS.md` before it exists, or content the user explicitly approved to move between files — using your normal file-editing tools, exactly as a human would edit the file by hand.\n- Verify the result by calling `uze agent context inspect` again after any change.\n\nYou MUST NOT:\n- Write, edit, or delete anything between `\u003C!-- uze:begin ... -->` \u002F `\u003C!-- uze:end ... -->` markers. They are what an earlier UZE left; `uze install` (or `uze agent context reconcile`) removes them, and one somebody edited by hand is theirs to keep or delete.\n- Invent your own markers, receipts, or a `CLAUDE.md` bridge: every harness reads `AGENTS.md` itself.\n- Run shell one-liners, `sed`, or scripts that touch `AGENTS.md`\u002F`CLAUDE.md`\u002F`GEMINI.md` as a substitute for `uze agent context reconcile`.\n- Implement your own version of inspect\u002Fplan\u002Freconcile logic (e.g. hand-computing what's drifted). Always ask the CLI; never infer state from a stale memory of a previous call.\n- Apply any `AGENTS.md` change without an explicit human confirmation first.\n- Silently overwrite or \"fix\" a `DRIFTED` state you see reported. Report it and ask.\n- Read the repository indiscriminately. Analysis is bounded (see below).\n\n## Step 1 — always start with the deterministic truth\n\nRun first, before saying anything about the project's context:\n\n```bash\nuze agent context inspect --format json\n```\n\nThis is read-only and safe to run at any time, including mid-conversation to\nre-check. Use its `portability` field to decide which flow below applies.\nDo not guess at file existence or drift state yourself — this is the one\nsource of truth.\n\n## Step 2 — branch on what inspect found\n\n### `PORTABLE`\n\nNothing to propose. Tell the user their context is already portable and\nsummarize `harnesses` briefly (who reads it natively, who is not detected\non this machine). Stop here unless they ask for something\nspecific.\n\n### Sections an earlier UZE wrote\n\nAn `AGENTS.md` may still carry blocks between `\u003C!-- uze:begin … -->`\nmarkers: a \"Concurrent work isolation\" or an \"Authoring plugins\" section\n(`project:…`), or a plugin's own instructions (`package:\u003Cplugin>:\u003Cmarket>:instructions`),\nand a `CLAUDE.md` may hold only an `@AGENTS.md` bridge between\n`instruction-bridge` markers. Earlier releases wrote them; nothing does any\nmore. A launched agent gets the workspace's instructions with its context,\nthe authoring verbs come with the `uze:author` Skill, and a plugin speaks to\nevery agent through an `always_on` rule. They are not the project's\ncontent: never fold them into a proposal. Report them in the inventory\n(`retired_regions` in `uze agent context inspect`), then `uze install` or\n`uze agent context reconcile` removes them in the main checkout, with the\nbridge and a `CLAUDE.md` that held nothing else. One someone edited by hand\nis reported and left for the person to keep or delete. If a plugin's block\nsaid something the project still wants, propose it as the person's own\ntext, or tell them the plugin's author can ship it as an `always_on` rule.\n\n### `NO_CONTEXT` (no AGENTS.md, no CLAUDE.md, no GEMINI.md)\n\nThis is the \"portable init\" flow — see **Flow A** below.\n\n### `VENDOR_LOCKED` (a vendor file has content, no AGENTS.md)\n\nThis is the \"extract the portable core\" flow — see **Flow B** below.\n\n### Rules kept for one harness, or hiding in `AGENTS.md`\n\nBeside the context files, look for rules that reach only one harness:\n`.claude\u002Frules\u002F`, `.cursor\u002Frules\u002F` (`.mdc`), `.windsurf\u002Frules\u002F` and the legacy\n`.agent\u002Frules\u002F`. Look too for `AGENTS.md` sections that only apply under one\npath (\"In `src\u002Fui\u002F`…\", \"For migrations…\"). Report each in the inventory and\npropose moving it to `.agents\u002Frules\u002F`, which every harness reads, through the\n`uze:rules` Skill. The frontmatter maps mechanically: `paths` or `globs`\nbecomes `trigger: glob` (a pattern naming a directory gains a leading `**\u002F`), `alwaysApply: true` becomes `trigger: always_on`,\nand a `description` alone becomes `trigger: model_decision`. A pattern\noutside the supported dialect (braces, character classes, a leading `\u002F`) is\nflagged for the person to decide, never rewritten. A move is user-owned\ncontent: propose it, wait for agreement, and finish with\n`uze agent rules check`.\n\n## Flow A — no context exists yet (\"portable init\")\n\nBounded analysis only. Read, in this order, stopping as soon as you have\nenough evidence — do not read the whole source tree:\n\n1. Package manifests at the root: `package.json`, `Cargo.toml`,\n   `pyproject.toml`, `go.mod`, `Gemfile`, etc. — language, package manager,\n   declared scripts.\n2. `README.md` — project purpose, stated build\u002Ftest\u002Frun instructions.\n3. Top-level CI config (`.github\u002Fworkflows\u002F*`, etc.) — the commands a human\n   already trusts enough to gate merges with; strong signal for what\n   \"tests pass\" and \"lint passes\" actually mean here.\n4. Existing `docs\u002F`\u002F`ARCHITECTURE.md`-shaped files, if present — architecture\n   conventions worth preserving.\n5. Top-level directory structure (listing, not content) — monorepo vs.\n   single package, workspace layout.\n6. The rules declared plugins already copied into `.agents\u002Frules\u002F`\n   (`uze status`): what they say every agent hears anyway, so the proposal\n   need not repeat it.\n\nDo not open individual source files looking for conventions unless a signal\nabove is genuinely ambiguous and one targeted read would resolve it.\n\nDraft an `AGENTS.md` proposal: a short project overview, the real\nbuild\u002Ftest\u002Flint\u002Fformat commands (from evidence, not assumption), and\nengineering conventions you found actual evidence for — not generic\nboilerplate. Quality bar: comparable to a mature harness's own `\u002Finit`, but\nthe output is portable `AGENTS.md` content, never a vendor-proprietary file.\n\nThen go to Step 3 (present the dry-run) — never write anything yet.\n\n## Flow B — a vendor file exists, no AGENTS.md (\"extract the portable core\")\n\nRead the existing vendor file(s) (`CLAUDE.md` and\u002For `GEMINI.md`). For each\ninstruction in them, classify it:\n\n- **Portable candidate** — build\u002Ftest\u002Flint commands, project conventions,\n  architecture notes, anything that would help *any* coding agent working\n  on this repo. This is what you propose moving into `AGENTS.md`.\n- **Vendor-specific** — instructions that only make sense for that one\n  harness (e.g. \"use Claude subagents for code review,\" a\n  vendor-specific workflow). This stays in the vendor file, unchanged.\n- **Ambiguous** — you are not confident which bucket it belongs in. Ask\n  the user (Step 3 groups these into one question, doesn't ask one at a\n  time).\n\n**Never do `cp CLAUDE.md AGENTS.md`.** Never move vendor-specific content.\nNever delete anything from the vendor file — content that stays\nvendor-specific stays exactly where it is; only the portable subset gets\nproposed as new `AGENTS.md` content.\n\nIf both `CLAUDE.md` and `GEMINI.md` exist with different content and no\n`AGENTS.md`: read both, propose one merged portable core, and be explicit\nin the dry-run about which harness each vendor-specific fragment came from\n— do not silently decide the two files \"mean the same thing.\"\n\n## Step 3 — present the dry-run (two things, never merged into one)\n\nShow these as two clearly separate sections. The user needs to tell them\napart at a glance:\n\n**1. Semantic proposal** — the content you are proposing, in full or as a\nsubstantive summary. This is your reasoning, not yet applied.\n\n**2. Deterministic UZE plan** — the output of `uze agent context plan`, verbatim\nor lightly formatted: which blocks an earlier release wrote would be\nremoved, and any that would be kept because someone edited them or their\nmarkers are malformed (flag those prominently: they need the user's\ndecision, not yours). A project with none has nothing in this section.\n\nExample shape:\n\n```\nUZE analyzed this project.\n\nDetected\n  Rust workspace, 5 crates\n  cargo test \u002F cargo clippy \u002F cargo fmt (from CI config)\n\nNo portable project context exists.\n\nProposed AGENTS.md content\n  [the actual drafted content, or a clear summary of it]\n\nUZE plan (uze agent context plan)\n  nothing to remove\n  Claude Code native, reads AGENTS.md beside its own CLAUDE.md\n  Codex       native\n  OpenCode    native\n  Antigravity native (reads AGENTS.md and GEMINI.md directly)\n\nApply?\n```\n\nGroup any ambiguous-classification questions from Flow B here too, in the\nsame turn — do not turn this into a multi-round wizard.\n\n## Step 4 — apply only after explicit confirmation\n\n1. Write the approved **user-owned** content (new `AGENTS.md` prose, or\n   vendor-specific fragments moved within a vendor file) with your normal\n   file tools: it is content the user owns.\n2. If the plan listed blocks an earlier release wrote, run\n   `uze agent context reconcile --format json` (or `uze install`) to remove\n   them. Never cut them out by hand, even if you believe you know the exact\n   bytes the removal would leave.\n3. If reconcile reports a block it kept (edited by hand, or malformed\n   markers), stop and report it. Do not retry with a workaround.\n\n## Step 5 — verify and report\n\nRun `uze agent context inspect --format json` one more time and report the real,\ncurrent, verified state — never the state you expect based on what you\njust did:\n\n```\nPortable context created.\n\nAGENTS.md   healthy\nClaude      native\nCodex       native\nOpenCode    native\nAntigravity native\n\nPortability: PORTABLE\n```\n\n## Already-healthy and other steady states\n\n- **Healthy context** (Step 2's `PORTABLE` branch): say so plainly, do not\n  manufacture work. \"Project context is healthy and portable. No changes\n  required.\"\n- **A block an earlier UZE wrote, edited by hand**: `uze agent context plan`\n  reports it as kept. Say where it is and ask the user whether to keep its\n  text as their own (they delete the markers) or delete the whole block.\n  Either is a human decision about *content*, never yours.\n- **A harness not detected on this machine**: `uze agent context inspect` reports\n  this as `NotDetected`, not a gap. Mention it factually if relevant, don't\n  treat it as something to fix.\n\n## Asking questions\n\n- Infer from strong evidence (a `package.json` `\"scripts\".test` field is\n  strong evidence for the test command; a lone ambiguous comment is not).\n- Ask when there's a genuine choice a human should make — which vendor\n  fragment is portable, which of two plausible test commands is the real\n  one, whether to keep the text of an edited block an earlier UZE wrote.\n- Never ask about something you already have strong evidence for.\n- Group questions into the Step 3 dry-run turn. Do not run a multi-step\n  wizard.\n","> This skill runs `uze`. If `uze` is not on `PATH`, UZE is not installed here and nothing below applies.\n\n# UZE — agentic context orchestrator\n\nYou are UZE's own Skill: the agentic layer of a two-layer system.\n\n```\nYou (this Skill)              UZE Context Manager (the `uze` CLI)\n  reasoning                     inspect   — read-only, ground truth\n  semantic analysis             plan      — read-only, what reconcile would remove\n  proposal                      reconcile — removes what an earlier UZE wrote\n  conversation\n```\n\n**The boundary is absolute.** `AGENTS.md` is the project's own: `uze` writes\nnothing into it. The one thing `uze` still does to it is take out the\n`\u003C!-- uze:begin … -->` blocks an earlier release wrote, and the one-line\n`CLAUDE.md` bridge that went with them. You reason, propose, and converse;\nthe content you draft is the human's, written with your normal file tools\nonce they agree, and the old blocks go only through `uze`.\n\n## Hard boundaries — read before doing anything\n\nYou MAY:\n- Run `uze agent context inspect`, `uze agent context plan`, `uze agent context reconcile` (all accept `--format json`).\n- Read files: `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`, package manifests, README, CI config, docs.\n- Analyze the project's stack, build\u002Ftest\u002Flint commands, structure, and conventions.\n- Ask the user questions and propose content.\n- Write **user-owned, non-managed content** — e.g. drafting new prose for `AGENTS.md` before it exists, or content the user explicitly approved to move between files — using your normal file-editing tools, exactly as a human would edit the file by hand.\n- Verify the result by calling `uze agent context inspect` again after any change.\n\nYou MUST NOT:\n- Write, edit, or delete anything between `\u003C!-- uze:begin ... -->` \u002F `\u003C!-- uze:end ... -->` markers. They are what an earlier UZE left; `uze install` (or `uze agent context reconcile`) removes them, and one somebody edited by hand is theirs to keep or delete.\n- Invent your own markers, receipts, or a `CLAUDE.md` bridge: every harness reads `AGENTS.md` itself.\n- Run shell one-liners, `sed`, or scripts that touch `AGENTS.md`\u002F`CLAUDE.md`\u002F`GEMINI.md` as a substitute for `uze agent context reconcile`.\n- Implement your own version of inspect\u002Fplan\u002Freconcile logic (e.g. hand-computing what's drifted). Always ask the CLI; never infer state from a stale memory of a previous call.\n- Apply any `AGENTS.md` change without an explicit human confirmation first.\n- Silently overwrite or \"fix\" a `DRIFTED` state you see reported. Report it and ask.\n- Read the repository indiscriminately. Analysis is bounded (see below).\n\n## Step 1 — always start with the deterministic truth\n\nRun first, before saying anything about the project's context:\n\n```bash\nuze agent context inspect --format json\n```\n\nThis is read-only and safe to run at any time, including mid-conversation to\nre-check. Use its `portability` field to decide which flow below applies.\nDo not guess at file existence or drift state yourself — this is the one\nsource of truth.\n\n## Step 2 — branch on what inspect found\n\n### `PORTABLE`\n\nNothing to propose. Tell the user their context is already portable and\nsummarize `harnesses` briefly (who reads it natively, who is not detected\non this machine). Stop here unless they ask for something\nspecific.\n\n### Sections an earlier UZE wrote\n\nAn `AGENTS.md` may still carry blocks between `\u003C!-- uze:begin … -->`\nmarkers: a \"Concurrent work isolation\" or an \"Authoring plugins\" section\n(`project:…`), or a plugin's own instructions (`package:\u003Cplugin>:\u003Cmarket>:instructions`),\nand a `CLAUDE.md` may hold only an `@AGENTS.md` bridge between\n`instruction-bridge` markers. Earlier releases wrote them; nothing does any\nmore. A launched agent gets the workspace's instructions with its context,\nthe authoring verbs come with the `uze:author` Skill, and a plugin speaks to\nevery agent through an `always_on` rule. They are not the project's\ncontent: never fold them into a proposal. Report them in the inventory\n(`retired_regions` in `uze agent context inspect`), then `uze install` or\n`uze agent context reconcile` removes them in the main checkout, with the\nbridge and a `CLAUDE.md` that held nothing else. One someone edited by hand\nis reported and left for the person to keep or delete. If a plugin's block\nsaid something the project still wants, propose it as the person's own\ntext, or tell them the plugin's author can ship it as an `always_on` rule.\n\n### `NO_CONTEXT` (no AGENTS.md, no CLAUDE.md, no GEMINI.md)\n\nThis is the \"portable init\" flow — see **Flow A** below.\n\n### `VENDOR_LOCKED` (a vendor file has content, no AGENTS.md)\n\nThis is the \"extract the portable core\" flow — see **Flow B** below.\n\n### Rules kept for one harness, or hiding in `AGENTS.md`\n\nBeside the context files, look for rules that reach only one harness:\n`.claude\u002Frules\u002F`, `.cursor\u002Frules\u002F` (`.mdc`), `.windsurf\u002Frules\u002F` and the legacy\n`.agent\u002Frules\u002F`. Look too for `AGENTS.md` sections that only apply under one\npath (\"In `src\u002Fui\u002F`…\", \"For migrations…\"). Report each in the inventory and\npropose moving it to `.agents\u002Frules\u002F`, which every harness reads, through the\n`uze:rules` Skill. The frontmatter maps mechanically: `paths` or `globs`\nbecomes `trigger: glob` (a pattern naming a directory gains a leading `**\u002F`), `alwaysApply: true` becomes `trigger: always_on`,\nand a `description` alone becomes `trigger: model_decision`. A pattern\noutside the supported dialect (braces, character classes, a leading `\u002F`) is\nflagged for the person to decide, never rewritten. A move is user-owned\ncontent: propose it, wait for agreement, and finish with\n`uze agent rules check`.\n\n## Flow A — no context exists yet (\"portable init\")\n\nBounded analysis only. Read, in this order, stopping as soon as you have\nenough evidence — do not read the whole source tree:\n\n1. Package manifests at the root: `package.json`, `Cargo.toml`,\n   `pyproject.toml`, `go.mod`, `Gemfile`, etc. — language, package manager,\n   declared scripts.\n2. `README.md` — project purpose, stated build\u002Ftest\u002Frun instructions.\n3. Top-level CI config (`.github\u002Fworkflows\u002F*`, etc.) — the commands a human\n   already trusts enough to gate merges with; strong signal for what\n   \"tests pass\" and \"lint passes\" actually mean here.\n4. Existing `docs\u002F`\u002F`ARCHITECTURE.md`-shaped files, if present — architecture\n   conventions worth preserving.\n5. Top-level directory structure (listing, not content) — monorepo vs.\n   single package, workspace layout.\n6. The rules declared plugins already copied into `.agents\u002Frules\u002F`\n   (`uze status`): what they say every agent hears anyway, so the proposal\n   need not repeat it.\n\nDo not open individual source files looking for conventions unless a signal\nabove is genuinely ambiguous and one targeted read would resolve it.\n\nDraft an `AGENTS.md` proposal: a short project overview, the real\nbuild\u002Ftest\u002Flint\u002Fformat commands (from evidence, not assumption), and\nengineering conventions you found actual evidence for — not generic\nboilerplate. Quality bar: comparable to a mature harness's own `\u002Finit`, but\nthe output is portable `AGENTS.md` content, never a vendor-proprietary file.\n\nThen go to Step 3 (present the dry-run) — never write anything yet.\n\n## Flow B — a vendor file exists, no AGENTS.md (\"extract the portable core\")\n\nRead the existing vendor file(s) (`CLAUDE.md` and\u002For `GEMINI.md`). For each\ninstruction in them, classify it:\n\n- **Portable candidate** — build\u002Ftest\u002Flint commands, project conventions,\n  architecture notes, anything that would help *any* coding agent working\n  on this repo. This is what you propose moving into `AGENTS.md`.\n- **Vendor-specific** — instructions that only make sense for that one\n  harness (e.g. \"use Claude subagents for code review,\" a\n  vendor-specific workflow). This stays in the vendor file, unchanged.\n- **Ambiguous** — you are not confident which bucket it belongs in. Ask\n  the user (Step 3 groups these into one question, doesn't ask one at a\n  time).\n\n**Never do `cp CLAUDE.md AGENTS.md`.** Never move vendor-specific content.\nNever delete anything from the vendor file — content that stays\nvendor-specific stays exactly where it is; only the portable subset gets\nproposed as new `AGENTS.md` content.\n\nIf both `CLAUDE.md` and `GEMINI.md` exist with different content and no\n`AGENTS.md`: read both, propose one merged portable core, and be explicit\nin the dry-run about which harness each vendor-specific fragment came from\n— do not silently decide the two files \"mean the same thing.\"\n\n## Step 3 — present the dry-run (two things, never merged into one)\n\nShow these as two clearly separate sections. The user needs to tell them\napart at a glance:\n\n**1. Semantic proposal** — the content you are proposing, in full or as a\nsubstantive summary. This is your reasoning, not yet applied.\n\n**2. Deterministic UZE plan** — the output of `uze agent context plan`, verbatim\nor lightly formatted: which blocks an earlier release wrote would be\nremoved, and any that would be kept because someone edited them or their\nmarkers are malformed (flag those prominently: they need the user's\ndecision, not yours). A project with none has nothing in this section.\n\nExample shape:\n\n```\nUZE analyzed this project.\n\nDetected\n  Rust workspace, 5 crates\n  cargo test \u002F cargo clippy \u002F cargo fmt (from CI config)\n\nNo portable project context exists.\n\nProposed AGENTS.md content\n  [the actual drafted content, or a clear summary of it]\n\nUZE plan (uze agent context plan)\n  nothing to remove\n  Claude Code native, reads AGENTS.md beside its own CLAUDE.md\n  Codex       native\n  OpenCode    native\n  Antigravity native (reads AGENTS.md and GEMINI.md directly)\n\nApply?\n```\n\nGroup any ambiguous-classification questions from Flow B here too, in the\nsame turn — do not turn this into a multi-round wizard.\n\n## Step 4 — apply only after explicit confirmation\n\n1. Write the approved **user-owned** content (new `AGENTS.md` prose, or\n   vendor-specific fragments moved within a vendor file) with your normal\n   file tools: it is content the user owns.\n2. If the plan listed blocks an earlier release wrote, run\n   `uze agent context reconcile --format json` (or `uze install`) to remove\n   them. Never cut them out by hand, even if you believe you know the exact\n   bytes the removal would leave.\n3. If reconcile reports a block it kept (edited by hand, or malformed\n   markers), stop and report it. Do not retry with a workaround.\n\n## Step 5 — verify and report\n\nRun `uze agent context inspect --format json` one more time and report the real,\ncurrent, verified state — never the state you expect based on what you\njust did:\n\n```\nPortable context created.\n\nAGENTS.md   healthy\nClaude      native\nCodex       native\nOpenCode    native\nAntigravity native\n\nPortability: PORTABLE\n```\n\n## Already-healthy and other steady states\n\n- **Healthy context** (Step 2's `PORTABLE` branch): say so plainly, do not\n  manufacture work. \"Project context is healthy and portable. No changes\n  required.\"\n- **A block an earlier UZE wrote, edited by hand**: `uze agent context plan`\n  reports it as kept. Say where it is and ask the user whether to keep its\n  text as their own (they delete the markers) or delete the whole block.\n  Either is a human decision about *content*, never yours.\n- **A harness not detected on this machine**: `uze agent context inspect` reports\n  this as `NotDetected`, not a gap. Mention it factually if relevant, don't\n  treat it as something to fix.\n\n## Asking questions\n\n- Infer from strong evidence (a `package.json` `\"scripts\".test` field is\n  strong evidence for the test command; a lone ambiguous comment is not).\n- Ask when there's a genuine choice a human should make — which vendor\n  fragment is portable, which of two plausible test commands is the real\n  one, whether to keep the text of an edited block an earlier UZE wrote.\n- Never ask about something you already have strong evidence for.\n- Group questions into the Step 3 dry-run turn. Do not run a multi-step\n  wizard.\n",{"name":5,"description":7},"main","e2cfbb55eec2aa007135f9a2256b73f7170a3b9e","plugins\u002Fuze\u002Fskills\u002Finit",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":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add uze-sh\u002Fuze --skill init",{"tool":39,"command":40},"bunx","bunx skills add uze-sh\u002Fuze --skill init",{"tool":42,"command":43},"pnpm","pnpm dlx skills add uze-sh\u002Fuze --skill init","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fuze-sh\u002Fuze\u002Fblob\u002Fmain\u002Fplugins\u002Fuze\u002Fskills\u002Finit\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"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":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"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":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"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":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"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":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"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":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"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",[],[]]