[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75YRM7AA784YW5818EM0AXD":3,"skill-same:01M75YRM7AA784YW5818EM0AXD":52},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M75YRM7AA784YW5818EM0AXD","setup",null,"Install and configure atl authentication, backends, and mirror defaults. USE WHEN the user explicitly requests install, authentication, setup repair, or $setup. DO NOT USE WHEN handling normal Jira, Confluence, search, reporting, or mirror work; explicit-only.","isukharev\u002Fatl","isukharev","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F4347261?v=4","01M75YQ67178RKPTQQKHF1V28G","Ivan Sukharev","01M75YQ68359TNKPGE26PY1Q33",7,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fplugins\u002Fatl\u002Fskills\u002Fsetup\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: setup\ndescription: Install and configure atl authentication, backends, and mirror defaults. USE WHEN the user explicitly requests install, authentication, setup repair, or $setup. DO NOT USE WHEN handling normal Jira, Confluence, search, reporting, or mirror work; explicit-only.\ndisable-model-invocation: true\nallowed-tools: Bash(command -v atl) Bash(atl version) Bash(brew install *) Bash(brew upgrade *) Bash(curl -fsSL https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Freleases\u002Flatest\u002Fdownload\u002Finstall.sh | sh) Bash(go install *) Bash(go env *) Bash(echo *) Bash(atl config show) Bash(atl config set *) Bash(atl compatibility status *) Bash(atl compatibility pin *) Bash(atl compatibility clear *) Bash(atl auth status) Bash(atl auth login *) Bash(atl conf search *) Bash(atl conf status *) Bash(atl jira fields *) Bash(atl jira status *)\n---\n\u003C!-- Generated from skills-src\u002Fsetup\u002FSKILL.md — edit the source and run 'make gen-plugins'. -->\n\n# Set up the atl CLI\n\nGet the user from zero to ready: install the `atl` binary, point it at their Confluence\u002FJira,\nauthenticate, and agree on where the local mirror lives. Work through the steps in order; skip a\nstep only if its check shows it is already done.\n\nIf the user wants an investigation-only agent\u002FCI profile, finish all required\nURL\u002Fauth setup first, then offer `atl config set safety.read_only true` as the\nlast step. Explain that persistent mode intentionally blocks later `config set`\nuntil a human edits\u002Fremoves `read_only` in the owner-only global config file.\n\nWhen setup commands fail, use JSON `kind` and `remediation` rather than parsing\nthe backend message. The schema-v1 `recovery` action is typed routing guidance,\nnot permission to mutate configuration; its `retry_safe` flag concerns only an\nexact replay. `configuration_error` means complete local setup;\n`authentication_failed` means replace\u002Fre-enter the rejected credential.\nIf `config.json` itself is malformed, `atl version`, help\u002Fcompletion, offline\nprofile\u002Fauth diagnostics, and local-only `conf status` \u002F `jira status` still\nwork; use their evidence, then have the human repair the owner-only file. Do not\nadd `--remote`, attempt other online reads, or perform mutations.\n\nInvocation: install\u002Fenable the atl plugin in Codex, then run this skill from `\u002Fskills` or with `$setup`.\n\n## 1. Detect an existing install\n\n```bash\ncommand -v atl && atl version\n```\n\nIf `atl` is found, report `version`, the full `commit`, and `build_state`, then\nskip to step 3. `clean` identifies a supported build from an unchanged source\ncheckout; `dirty` means local source changes were present; `unknown` is valid\nfor an unstamped build whose Go VCS metadata is unavailable. These are\ndiagnostics, not signature or update-trust evidence. Otherwise continue.\n\n## 2. Install the binary\n\nPick the method that fits the platform:\n\n**macOS (or Linuxbrew) with Homebrew — preferred there** (handles `PATH` for you):\n\n```bash\nbrew install isukharev\u002Ftap\u002Fatl\n```\n\nThe Homebrew launcher sets `ATL_NO_UPDATE=1`: Homebrew is the only owner of\nthat installation, so later upgrades use `brew upgrade atl`. Direct\ninstaller\u002Frelease binaries retain signed self-update unless their environment\nsets `ATL_NO_UPDATE`. Remote work for a due check has one five-second total\nstartup budget.\n\n**Linux, or macOS without Homebrew** — prebuilt, SHA-256 verified, installs to `~\u002F.local\u002Fbin\u002Fatl`:\n\n```bash\ncurl -fsSL https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Freleases\u002Flatest\u002Fdownload\u002Finstall.sh | sh\n```\n\nBoth the `brew` and `curl | sh` paths are network installs that Codex will ask the user to\napprove. That prompt is expected; do not try to bypass it.\n\n**Fallback**, if the above fail and Go is installed (the `main` package lives in `cmd\u002Fatl`, so the\nmodule path must end in `\u002Fcmd\u002Fatl`):\n\n```bash\ngo install github.com\u002Fisukharev\u002Fatl\u002Fcmd\u002Fatl@latest\n```\n\nAfter installing, confirm `atl` resolves **in this shell** (the `curl`\u002F`go install` paths do not add\nit to `PATH` automatically — `brew` does):\n\n```bash\ncommand -v atl && atl version || echo 'atl is not on PATH — add ~\u002F.local\u002Fbin (or $(go env GOBIN)) to PATH'\n```\n\nIf it is not on `PATH`, give the user the exact line to add to their shell profile (e.g.\n`export PATH=\"$HOME\u002F.local\u002Fbin:$PATH\"`); do not edit their profile silently. **Do not continue to\nstep 3 until `atl version` prints a version in the current session** — otherwise every later step\nfails with \"command not found\".\n\n## 3. Configure backend URLs\n\nAsk the user for their Confluence and Jira base URLs. An unambiguous hostname (optionally with a\nport and base path) may omit `https:\u002F\u002F`; `atl config set` stores it as HTTPS. An explicit\n`http:\u002F\u002F` URL remains refused unless the user explicitly enables the documented insecure-transport\noverride. Set whichever service they use — both is typical:\n\n```bash\natl config set --confluence-url https:\u002F\u002Fconfluence.example.com --jira-url https:\u002F\u002Fjira.example.com\natl config show\n```\n\n`config show` prints `confluence_url`, `jira_url`, `update_base_url`, a `mirror` hint block\nwith the recommended `~\u002F.atl\u002F\u003Cworkspace>\u002F` root plus active `ATL_MIRROR_ROOT` when set, and the\neffective `render` block with `render_provenance` (see below). Its shared\n`render.display_time_zone` is an IANA presentation zone (deterministic `UTC` by default); it\nchanges human Markdown dates only, never JQL\u002FCQL or exact JSON\u002Fnative timestamps. A non-https URL for a non-loopback\nhost is rejected at set time.\n\nIf an HTTPS backend uses a private CA, set only that backend's reviewed PEM\nbundle before authentication. The path is persisted but never printed:\n\n```bash\natl config set transport.confluence.ca_bundle \u002Fpath\u002Fto\u002Fconfluence-ca.pem\natl config set transport.jira.ca_bundle \u002Fpath\u002Fto\u002Fjira-ca.pem\natl doctor\n```\n\nDo not request or configure a client certificate or private key; ATL supports\nadditional trust roots only.\n\n**Render config layer (presentation-only).** `config set` also takes a positional dotted render key\nthat tunes the derived `.md` view — `render.{jira,confluence}.{profile,include,exclude}` (profile is\n`minimal`|`default`|`full`) plus Jira-only `custom_fields` (comma-separated), typed\n`field_views` (JSON descriptor array), and `epic_field`; Confluence has a\nclosed read-only `page_fields` JSON descriptor array:\n\n```bash\natl config set render.display_time_zone Europe\u002FBerlin # global human-readable Markdown zone\natl config set render.jira.profile full            # global (~\u002F.config\u002Fatl\u002Fconfig.json)\natl config set --local render.confluence.profile minimal   # per-mirror \u003Croot>\u002F.atl\u002Fconfig.json\natl config set --local render.confluence.page_fields '[{\"id\":\"title\"},{\"id\":\"updated\",\"format\":\"date\"}]'\natl config set render.confluence.jira_macros off # global-only: never execute page JQL\natl config set --local render.jira.field_views '[{\"id\":\"customfield_10003\",\"label\":\"Risk Notes\",\"placement\":\"section\",\"format\":\"jira_wiki\",\"editable\":true}]'\n```\n\n`--local` writes a per-mirror `.atl\u002Fconfig.json` (nearest `.atl` walking up from cwd, or `--into ROOT`).\nIt is a **security boundary**: a local file may carry render keys only — backend\u002Fupdate URLs are\nglobal\u002Fenv-only so a shared or checked-out mirror can never redirect where the PAT is sent. `config\nset --local` refuses a URL flag (exit 2); at read time any forbidden\u002Funknown key in a local file is\nwarned about on stderr and ignored. Precedence is local > global > default. Pull\u002Frender consume the\neffective settings and record the resolved view in `.atl\u002Fstate.json` for apply affinity.\n\n## 4. Authenticate\n\n`atl` uses a per-service Personal Access Token (PAT). **Never put a PAT on the command line** —\n`auth login` reads it from a no-echo prompt, piped stdin, or a file. Recommend a least-privilege,\ntask-scoped token.\n\nInteractive (the agent runs this; the user types the token at the hidden prompt):\n\n```bash\natl auth login --service confluence\natl auth login --service jira\n```\n\nOr from a file: `atl auth login --service jira --from-file .\u002Ftoken.txt` (then delete the file).\nOr via environment for CI\u002Fagent sessions: `ATL_CONFLUENCE_PAT` \u002F `ATL_JIRA_PAT`.\n\nFor rare direct REST fallback calls that `atl` does not cover yet, keep the PAT out of argv\u002Flogs:\nuse env vars, turn off shell tracing, and feed curl config\u002Fheaders through stdin instead of\n`curl -H \"Authorization: Bearer $TOKEN\"`.\n\nVerify (this never prints the token, only where it resolves from):\n\n```bash\natl auth status\natl doctor\n```\n\n`atl doctor` is the preferred share-safe setup report. It is offline by default\nand emits no configured URL\u002Fhostname, path, identity, token, or mirror content.\nIf it exits `8`, inspect its emitted `problems[]`; do not paste `config show`\ninto public chat or an issue.\n\n**Optional exact compatibility provider.** Do not enable one during ordinary\nsetup. When the owner explicitly needs a provider-specific Data Center\nworkflow, inspect `atl compatibility status`, review the exact owner-pinned\nversion\u002Fbuild, then run `compatibility pin` and `compatibility status --remote`.\nA nearby patch is unsupported; never substitute a version range or arbitrary\nREST fallback. `pin` and `clear` are persistent local mutations and require\nexplicit write authority.\n\n## 5. Agree on the mirror directory\n\n`atl` mirrors pages\u002Fissues to disk. Keep the mirror **out of the user's code repository** so it is\nfully greppable by the agent and never committed into their project's git history.\n\n- **Recommended convention:** `~\u002F.atl\u002F\u003Cworkspace>\u002F`, where `\u003Cworkspace>` is a\n  meaningful name (the code repo's basename or the Confluence space key).\n  Example: `~\u002F.atl\u002Fpayments-service\u002F`. This is not the pull\u002Fwrite fallback:\n  without `ATL_MIRROR_ROOT` or `--into`, Confluence pull uses `mirror` and Jira\n  pull uses `mirror-jira`; status\u002Fsnapshot first detect a nearest initialized\n  `.atl` as described below.\n- **Fix it once with `ATL_MIRROR_ROOT`** so Confluence\u002FJira pull plus\n  status\u002Fsnapshot inspection default to the same place without re-passing the\n  root every time. Status\u002Fsnapshot also accept positional `[DIR]` or `--into`\n  (never both), then try the nearest initialized `.atl` before the service\n  fallback. Record the environment setting where later sessions will pick it up —\n  either export it in the shell profile, or add a line to the project's `AGENTS.md`:\n  `atl mirror lives at ~\u002F.atl\u002F\u003Cworkspace>\u002F (export ATL_MIRROR_ROOT=~\u002F.atl\u002F\u003Cworkspace>\u002F)`.\n- An explicit `--into \u003Cdir>` still overrides `ATL_MIRROR_ROOT`. Inspection\n  requires an initialized `.atl` root and returns exit 4 before config\u002Fnetwork\n  when it is absent. `conf push` does not read the env\n  var — it finds the mirror root by walking up from the target file to the nearest `.atl`, so as long\n  as you push files from inside that same root it lines up automatically.\n\n(See the `atl` orientation skill's workflow reference for the full rationale and the in-repo \u002F\nscratch alternatives.)\n\n## 6. Smoke test\n\nConfirm auth + connectivity with a cheap read:\n\n```bash\natl doctor --remote\natl conf search --cql 'type = page' --limit 1   # if they use Confluence\natl jira fields --summary-only                   # if they use Jira\n```\n\n`doctor --remote` makes one single-attempt product\u002Fversion GET per ready\nservice. Only when the Confluence version route returns `404` may it add one\nbodyless reachability HEAD; that proves REST availability but leaves\ncompatibility unverified. It performs no search, page\u002Fissue body read, identity\nread, or write. The following bounded service read proves a useful\npermission\u002Fdata route.\n`atl` prints JSON by default. A healthy doctor result plus a clean, complete\nservice result means technical setup is complete. Offer one concrete next route\ninstead of loading the full command reference: a bounded focused read, a pull\ninto the agreed mirror followed by local status\u002Fdiff, or a non-writing preview\nfor a user-selected change. Use the `confluence` or `jira` skill for that route.\nDo not perform a write merely to prove setup.\n\nOffer the separate explicit `onboarding` skill if they want atl to learn their\nrecurring workflow, approved field\u002Fschema facts, render defaults, and common\nselectors. Do not run that interview or inspect sample content unless they opt\nin; technical setup remains complete without it.\n\nThe installed Claude Code\u002FCodex plugin also bundles `atl mcp serve`, a typed\nread-only evidence surface. Remote tools use the same configured host-scoped\ncredentials, while the two offline mirror snapshot tools use only an explicit\n`ATL_MIRROR_ROOT`; neither configuration is copied into plugin files. The\nsurface becomes available after starting a new agent session with `atl` on\n`PATH`. Its absence does not make CLI setup incomplete; use the CLI fallback and\nreport a plugin\u002Fbinary version skew instead of inventing raw REST calls.\nStandalone clients that need only one reviewed boundary may launch\n`atl mcp serve --service jira`, `--service confluence`, or\n`--service offline`. The default plugin launch remains the full read-only\ninventory; do not turn the profile flag into a model-selected arbitrary\nallowlist.\n\n## Version skew\n\nPlugin and binary release under one version number. If a command documented by the skills is\nrejected as unknown (exit 2), compare `atl version` with the plugin version and update the lagging\nside: a direct-install binary self-updates on its next run, while a Homebrew binary\nuses `brew upgrade atl`. Refresh the lagging plugin with the platform-specific flow:\n\nRun `codex plugin marketplace upgrade atl --json`. If it succeeds, run `codex plugin add atl@atl --json`. Then start a new Codex chat or CLI session before retrying.\n\n## Exit codes (so you can react)\n\n`2` usage · `3` auth (the server **rejected** the token → re-run step 4 with a valid PAT) ·\n`4` not-found · `5` version-conflict · `6` forbidden (token lacks permission) ·\n`7` not configured (backend URL or PAT **not set** yet → finish step 3\u002F4). Anything else is `1`.\n","\u003C!-- Generated from skills-src\u002Fsetup\u002FSKILL.md — edit the source and run 'make gen-plugins'. -->\n\n# Set up the atl CLI\n\nGet the user from zero to ready: install the `atl` binary, point it at their Confluence\u002FJira,\nauthenticate, and agree on where the local mirror lives. Work through the steps in order; skip a\nstep only if its check shows it is already done.\n\nIf the user wants an investigation-only agent\u002FCI profile, finish all required\nURL\u002Fauth setup first, then offer `atl config set safety.read_only true` as the\nlast step. Explain that persistent mode intentionally blocks later `config set`\nuntil a human edits\u002Fremoves `read_only` in the owner-only global config file.\n\nWhen setup commands fail, use JSON `kind` and `remediation` rather than parsing\nthe backend message. The schema-v1 `recovery` action is typed routing guidance,\nnot permission to mutate configuration; its `retry_safe` flag concerns only an\nexact replay. `configuration_error` means complete local setup;\n`authentication_failed` means replace\u002Fre-enter the rejected credential.\nIf `config.json` itself is malformed, `atl version`, help\u002Fcompletion, offline\nprofile\u002Fauth diagnostics, and local-only `conf status` \u002F `jira status` still\nwork; use their evidence, then have the human repair the owner-only file. Do not\nadd `--remote`, attempt other online reads, or perform mutations.\n\nInvocation: install\u002Fenable the atl plugin in Codex, then run this skill from `\u002Fskills` or with `$setup`.\n\n## 1. Detect an existing install\n\n```bash\ncommand -v atl && atl version\n```\n\nIf `atl` is found, report `version`, the full `commit`, and `build_state`, then\nskip to step 3. `clean` identifies a supported build from an unchanged source\ncheckout; `dirty` means local source changes were present; `unknown` is valid\nfor an unstamped build whose Go VCS metadata is unavailable. These are\ndiagnostics, not signature or update-trust evidence. Otherwise continue.\n\n## 2. Install the binary\n\nPick the method that fits the platform:\n\n**macOS (or Linuxbrew) with Homebrew — preferred there** (handles `PATH` for you):\n\n```bash\nbrew install isukharev\u002Ftap\u002Fatl\n```\n\nThe Homebrew launcher sets `ATL_NO_UPDATE=1`: Homebrew is the only owner of\nthat installation, so later upgrades use `brew upgrade atl`. Direct\ninstaller\u002Frelease binaries retain signed self-update unless their environment\nsets `ATL_NO_UPDATE`. Remote work for a due check has one five-second total\nstartup budget.\n\n**Linux, or macOS without Homebrew** — prebuilt, SHA-256 verified, installs to `~\u002F.local\u002Fbin\u002Fatl`:\n\n```bash\ncurl -fsSL https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Freleases\u002Flatest\u002Fdownload\u002Finstall.sh | sh\n```\n\nBoth the `brew` and `curl | sh` paths are network installs that Codex will ask the user to\napprove. That prompt is expected; do not try to bypass it.\n\n**Fallback**, if the above fail and Go is installed (the `main` package lives in `cmd\u002Fatl`, so the\nmodule path must end in `\u002Fcmd\u002Fatl`):\n\n```bash\ngo install github.com\u002Fisukharev\u002Fatl\u002Fcmd\u002Fatl@latest\n```\n\nAfter installing, confirm `atl` resolves **in this shell** (the `curl`\u002F`go install` paths do not add\nit to `PATH` automatically — `brew` does):\n\n```bash\ncommand -v atl && atl version || echo 'atl is not on PATH — add ~\u002F.local\u002Fbin (or $(go env GOBIN)) to PATH'\n```\n\nIf it is not on `PATH`, give the user the exact line to add to their shell profile (e.g.\n`export PATH=\"$HOME\u002F.local\u002Fbin:$PATH\"`); do not edit their profile silently. **Do not continue to\nstep 3 until `atl version` prints a version in the current session** — otherwise every later step\nfails with \"command not found\".\n\n## 3. Configure backend URLs\n\nAsk the user for their Confluence and Jira base URLs. An unambiguous hostname (optionally with a\nport and base path) may omit `https:\u002F\u002F`; `atl config set` stores it as HTTPS. An explicit\n`http:\u002F\u002F` URL remains refused unless the user explicitly enables the documented insecure-transport\noverride. Set whichever service they use — both is typical:\n\n```bash\natl config set --confluence-url https:\u002F\u002Fconfluence.example.com --jira-url https:\u002F\u002Fjira.example.com\natl config show\n```\n\n`config show` prints `confluence_url`, `jira_url`, `update_base_url`, a `mirror` hint block\nwith the recommended `~\u002F.atl\u002F\u003Cworkspace>\u002F` root plus active `ATL_MIRROR_ROOT` when set, and the\neffective `render` block with `render_provenance` (see below). Its shared\n`render.display_time_zone` is an IANA presentation zone (deterministic `UTC` by default); it\nchanges human Markdown dates only, never JQL\u002FCQL or exact JSON\u002Fnative timestamps. A non-https URL for a non-loopback\nhost is rejected at set time.\n\nIf an HTTPS backend uses a private CA, set only that backend's reviewed PEM\nbundle before authentication. The path is persisted but never printed:\n\n```bash\natl config set transport.confluence.ca_bundle \u002Fpath\u002Fto\u002Fconfluence-ca.pem\natl config set transport.jira.ca_bundle \u002Fpath\u002Fto\u002Fjira-ca.pem\natl doctor\n```\n\nDo not request or configure a client certificate or private key; ATL supports\nadditional trust roots only.\n\n**Render config layer (presentation-only).** `config set` also takes a positional dotted render key\nthat tunes the derived `.md` view — `render.{jira,confluence}.{profile,include,exclude}` (profile is\n`minimal`|`default`|`full`) plus Jira-only `custom_fields` (comma-separated), typed\n`field_views` (JSON descriptor array), and `epic_field`; Confluence has a\nclosed read-only `page_fields` JSON descriptor array:\n\n```bash\natl config set render.display_time_zone Europe\u002FBerlin # global human-readable Markdown zone\natl config set render.jira.profile full            # global (~\u002F.config\u002Fatl\u002Fconfig.json)\natl config set --local render.confluence.profile minimal   # per-mirror \u003Croot>\u002F.atl\u002Fconfig.json\natl config set --local render.confluence.page_fields '[{\"id\":\"title\"},{\"id\":\"updated\",\"format\":\"date\"}]'\natl config set render.confluence.jira_macros off # global-only: never execute page JQL\natl config set --local render.jira.field_views '[{\"id\":\"customfield_10003\",\"label\":\"Risk Notes\",\"placement\":\"section\",\"format\":\"jira_wiki\",\"editable\":true}]'\n```\n\n`--local` writes a per-mirror `.atl\u002Fconfig.json` (nearest `.atl` walking up from cwd, or `--into ROOT`).\nIt is a **security boundary**: a local file may carry render keys only — backend\u002Fupdate URLs are\nglobal\u002Fenv-only so a shared or checked-out mirror can never redirect where the PAT is sent. `config\nset --local` refuses a URL flag (exit 2); at read time any forbidden\u002Funknown key in a local file is\nwarned about on stderr and ignored. Precedence is local > global > default. Pull\u002Frender consume the\neffective settings and record the resolved view in `.atl\u002Fstate.json` for apply affinity.\n\n## 4. Authenticate\n\n`atl` uses a per-service Personal Access Token (PAT). **Never put a PAT on the command line** —\n`auth login` reads it from a no-echo prompt, piped stdin, or a file. Recommend a least-privilege,\ntask-scoped token.\n\nInteractive (the agent runs this; the user types the token at the hidden prompt):\n\n```bash\natl auth login --service confluence\natl auth login --service jira\n```\n\nOr from a file: `atl auth login --service jira --from-file .\u002Ftoken.txt` (then delete the file).\nOr via environment for CI\u002Fagent sessions: `ATL_CONFLUENCE_PAT` \u002F `ATL_JIRA_PAT`.\n\nFor rare direct REST fallback calls that `atl` does not cover yet, keep the PAT out of argv\u002Flogs:\nuse env vars, turn off shell tracing, and feed curl config\u002Fheaders through stdin instead of\n`curl -H \"Authorization: Bearer $TOKEN\"`.\n\nVerify (this never prints the token, only where it resolves from):\n\n```bash\natl auth status\natl doctor\n```\n\n`atl doctor` is the preferred share-safe setup report. It is offline by default\nand emits no configured URL\u002Fhostname, path, identity, token, or mirror content.\nIf it exits `8`, inspect its emitted `problems[]`; do not paste `config show`\ninto public chat or an issue.\n\n**Optional exact compatibility provider.** Do not enable one during ordinary\nsetup. When the owner explicitly needs a provider-specific Data Center\nworkflow, inspect `atl compatibility status`, review the exact owner-pinned\nversion\u002Fbuild, then run `compatibility pin` and `compatibility status --remote`.\nA nearby patch is unsupported; never substitute a version range or arbitrary\nREST fallback. `pin` and `clear` are persistent local mutations and require\nexplicit write authority.\n\n## 5. Agree on the mirror directory\n\n`atl` mirrors pages\u002Fissues to disk. Keep the mirror **out of the user's code repository** so it is\nfully greppable by the agent and never committed into their project's git history.\n\n- **Recommended convention:** `~\u002F.atl\u002F\u003Cworkspace>\u002F`, where `\u003Cworkspace>` is a\n  meaningful name (the code repo's basename or the Confluence space key).\n  Example: `~\u002F.atl\u002Fpayments-service\u002F`. This is not the pull\u002Fwrite fallback:\n  without `ATL_MIRROR_ROOT` or `--into`, Confluence pull uses `mirror` and Jira\n  pull uses `mirror-jira`; status\u002Fsnapshot first detect a nearest initialized\n  `.atl` as described below.\n- **Fix it once with `ATL_MIRROR_ROOT`** so Confluence\u002FJira pull plus\n  status\u002Fsnapshot inspection default to the same place without re-passing the\n  root every time. Status\u002Fsnapshot also accept positional `[DIR]` or `--into`\n  (never both), then try the nearest initialized `.atl` before the service\n  fallback. Record the environment setting where later sessions will pick it up —\n  either export it in the shell profile, or add a line to the project's `AGENTS.md`:\n  `atl mirror lives at ~\u002F.atl\u002F\u003Cworkspace>\u002F (export ATL_MIRROR_ROOT=~\u002F.atl\u002F\u003Cworkspace>\u002F)`.\n- An explicit `--into \u003Cdir>` still overrides `ATL_MIRROR_ROOT`. Inspection\n  requires an initialized `.atl` root and returns exit 4 before config\u002Fnetwork\n  when it is absent. `conf push` does not read the env\n  var — it finds the mirror root by walking up from the target file to the nearest `.atl`, so as long\n  as you push files from inside that same root it lines up automatically.\n\n(See the `atl` orientation skill's workflow reference for the full rationale and the in-repo \u002F\nscratch alternatives.)\n\n## 6. Smoke test\n\nConfirm auth + connectivity with a cheap read:\n\n```bash\natl doctor --remote\natl conf search --cql 'type = page' --limit 1   # if they use Confluence\natl jira fields --summary-only                   # if they use Jira\n```\n\n`doctor --remote` makes one single-attempt product\u002Fversion GET per ready\nservice. Only when the Confluence version route returns `404` may it add one\nbodyless reachability HEAD; that proves REST availability but leaves\ncompatibility unverified. It performs no search, page\u002Fissue body read, identity\nread, or write. The following bounded service read proves a useful\npermission\u002Fdata route.\n`atl` prints JSON by default. A healthy doctor result plus a clean, complete\nservice result means technical setup is complete. Offer one concrete next route\ninstead of loading the full command reference: a bounded focused read, a pull\ninto the agreed mirror followed by local status\u002Fdiff, or a non-writing preview\nfor a user-selected change. Use the `confluence` or `jira` skill for that route.\nDo not perform a write merely to prove setup.\n\nOffer the separate explicit `onboarding` skill if they want atl to learn their\nrecurring workflow, approved field\u002Fschema facts, render defaults, and common\nselectors. Do not run that interview or inspect sample content unless they opt\nin; technical setup remains complete without it.\n\nThe installed Claude Code\u002FCodex plugin also bundles `atl mcp serve`, a typed\nread-only evidence surface. Remote tools use the same configured host-scoped\ncredentials, while the two offline mirror snapshot tools use only an explicit\n`ATL_MIRROR_ROOT`; neither configuration is copied into plugin files. The\nsurface becomes available after starting a new agent session with `atl` on\n`PATH`. Its absence does not make CLI setup incomplete; use the CLI fallback and\nreport a plugin\u002Fbinary version skew instead of inventing raw REST calls.\nStandalone clients that need only one reviewed boundary may launch\n`atl mcp serve --service jira`, `--service confluence`, or\n`--service offline`. The default plugin launch remains the full read-only\ninventory; do not turn the profile flag into a model-selected arbitrary\nallowlist.\n\n## Version skew\n\nPlugin and binary release under one version number. If a command documented by the skills is\nrejected as unknown (exit 2), compare `atl version` with the plugin version and update the lagging\nside: a direct-install binary self-updates on its next run, while a Homebrew binary\nuses `brew upgrade atl`. Refresh the lagging plugin with the platform-specific flow:\n\nRun `codex plugin marketplace upgrade atl --json`. If it succeeds, run `codex plugin add atl@atl --json`. Then start a new Codex chat or CLI session before retrying.\n\n## Exit codes (so you can react)\n\n`2` usage · `3` auth (the server **rejected** the token → re-run step 4 with a valid PAT) ·\n`4` not-found · `5` version-conflict · `6` forbidden (token lacks permission) ·\n`7` not configured (backend URL or PAT **not set** yet → finish step 3\u002F4). Anything else is `1`.\n",{"name":5,"description":7,"allowed-tools":24,"disable-model-invocation":25},"Bash(command -v atl) Bash(atl version) Bash(brew install *) Bash(brew upgrade *) Bash(curl -fsSL https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Freleases\u002Flatest\u002Fdownload\u002Finstall.sh | sh) Bash(go install *) Bash(go env *) Bash(echo *) Bash(atl config show) Bash(atl config set *) Bash(atl compatibility status *) Bash(atl compatibility pin *) Bash(atl compatibility clear *) Bash(atl auth status) Bash(atl auth login *) Bash(atl conf search *) Bash(atl conf status *) Bash(atl jira fields *) Bash(atl jira status *)",true,"main","c6ed32ed384aa5e9a3e86d4528f16dc13833ce66","plugins\u002Fatl\u002Fskills\u002Fsetup",1,"https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl","https:\u002F\u002Fgithub.com\u002Fisukharev","2026-09-14 23:56:52",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add isukharev\u002Fatl --skill setup",{"tool":39,"command":40},"bunx","bunx skills add isukharev\u002Fatl --skill setup",{"tool":42,"command":43},"pnpm","pnpm dlx skills add isukharev\u002Fatl --skill setup","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fplugins\u002Fatl\u002Fskills\u002Fsetup\u002FSKILL.md",[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"agents\u002Fopenai.yaml","REFERENCE","参考文档",218,"https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fplugins\u002Fatl\u002Fskills\u002Fsetup\u002Fagents\u002Fopenai.yaml",[53,60,67,74,81,87],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M75YVB9DFRMQ0P4SB1RWCH8D","triage-issue","Triage a bug against Jira with atl before proposing a ticket. USE WHEN an error or stack trace comes with duplicate, prior-fix, or file-a-bug intent. DO NOT USE WHEN debugging code only, inspecting a known issue, or creating an ordinary task.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Ftriage-issue\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M75YV7K5C8A8NV1SYHM413B7","status-report","Draft an audience-oriented Jira status report with atl. USE WHEN the user wants a weekly\u002Fdaily progress narrative, blocker summary, or period report. DO NOT USE WHEN the output is a visual dashboard, direct Jira query, knowledge search, or codebase summary.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Fstatus-report\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M75YV57J2G080TFCV6QYWK2B","sprint-dashboard","Build a visual read-only Jira sprint or board dashboard with atl. USE WHEN the user wants a standup view, WIP picture, or sprint snapshot. DO NOT USE WHEN the output is a narrative report, raw Jira list, or codebase dashboard.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Fsprint-dashboard\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M75YV2JNBA9PTQA06JJSZGW5","spec-to-backlog","Convert a specification into a reviewed Jira epic and backlog with atl. USE WHEN decomposing requirements, RFCs, or designs into linked Jira work. DO NOT USE WHEN the source is meeting notes, only a summary is wanted, or the task is a direct service operation.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Fspec-to-backlog\u002FSKILL.md",[],[],{"skillNo":82,"name":5,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M75YTWKKWHKGZ0XQACM7VXNR","Install and configure atl authentication, backends, and mirror defaults. USE WHEN the user explicitly requests install, authentication, setup repair, or \u002Fatl:setup. DO NOT USE WHEN handling normal Jira, Confluence, search, reporting, or mirror work; explicit-only.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Fsetup\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M75YTSX670SW6NSBA1HGAG65","search-knowledge","Answer from qualified Jira and Confluence evidence with atl. USE WHEN organizational knowledge is unknown or split across both services. DO NOT USE WHEN one exact resource is known, the task is a report\u002Fdashboard, or the answer belongs in the codebase.","https:\u002F\u002Fgithub.com\u002Fisukharev\u002Fatl\u002Fblob\u002Fmain\u002Fskills\u002Fsearch-knowledge\u002FSKILL.md",[],[]]