[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79J76RN9128RDAVRMJ368X0":3,"skill-same:01M79J76RN9128RDAVRMJ368X0":50},{"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":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M79J76RN9128RDAVRMJ368X0","cli-setup-verify",null,"Verify the Omnigent CLI's setup\u002Fonboarding flow, terminal UI\u002FUX, and critical user journeys in a completely isolated, reproducible loop. Drives the real `omnigent` binary through a PTY (pexpect) inside a throwaway OMNIGENT_CONFIG_HOME \u002F OMNIGENT_DATA_DIR sandbox that never touches the user's real ~\u002F.omnigent, captures ANSI-stripped frames for UX inspection, and proves a change is verifiable via a before→fix→after baseline diff. Load when developing or reviewing a CLI setup\u002Fonboarding\u002FREPL\u002Fpicker change (omnigent\u002Fcli.py, omnigent\u002Fonboarding\u002F*, omnigent\u002Frepl\u002F*, scripts\u002Finstall_oss.sh), reproducing a cold-start\u002Ffirst-run UX bug, or confirming a fix actually lands. Several agents can run it concurrently on separate worktrees.","McClain-Thiel\u002FOmniSci","McClain-Thiel","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F26270512?v=4","01M79J6KEGVM49NSQMSWABYZTQ","McClain Thiel","01M79J6KFBH6W8J65XXM22M7N0",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcli-setup-verify\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: cli-setup-verify\ndescription: Verify the Omnigent CLI's setup\u002Fonboarding flow, terminal UI\u002FUX, and critical user journeys in a completely isolated, reproducible loop. Drives the real `omnigent` binary through a PTY (pexpect) inside a throwaway OMNIGENT_CONFIG_HOME \u002F OMNIGENT_DATA_DIR sandbox that never touches the user's real ~\u002F.omnigent, captures ANSI-stripped frames for UX inspection, and proves a change is verifiable via a before→fix→after baseline diff. Load when developing or reviewing a CLI setup\u002Fonboarding\u002FREPL\u002Fpicker change (omnigent\u002Fcli.py, omnigent\u002Fonboarding\u002F*, omnigent\u002Frepl\u002F*, scripts\u002Finstall_oss.sh), reproducing a cold-start\u002Ffirst-run UX bug, or confirming a fix actually lands. Several agents can run it concurrently on separate worktrees.\n---\n\n# Verifying the Omnigent CLI setup & UX in a closed loop\n\nThe Omnigent CLI's first impression is: `curl | sh` → run `omnigent` → pick a\nmodel credential → start a session. This skill lets an agent **enter that flow,\nexamine the UI\u002FUX, and prove whether a change is verifiable** — without a\nbrowser, without real credentials, and **without ever touching the developer's\nreal `~\u002F.omnigent`**.\n\nThe engine is `verify_cli.py` (next to this file). It drives the real\n`omnigent` binary through a pseudo-terminal (`pexpect`) inside a throwaway\nsandbox, captures what renders, runs assertions, and prints one machine-readable\n`SUMMARY {json}` line.\n\n> **The whole point is a verifiable loop**, not a one-shot check:\n> 1. Run a scenario on the **unfixed** code → baseline (`--label before`).\n> 2. Make the change.\n> 3. Run the **same** scenario → `--label after`.\n> 4. Diff the two `SUMMARY` lines. A fix is \"verifiable\" only if a concrete\n>    check or note **flips** between the two runs. If it doesn't flip, you\n>    can't prove the fix did anything — go back to step 2.\n\n## Why this is safe (read first)\n\nThe real `~\u002F.omnigent` here can be **many GB** (chat DB, runner logs, native\nharness state). The sandbox isolates every write three ways:\n\n- **`HOME` is redirected into the sandbox by default.** This is the load-bearing\n  one. `OMNIGENT_CONFIG_HOME` \u002F `OMNIGENT_DATA_DIR` (`omnigent\u002Fcli.py`\n  `_CONFIG_HOME_ENV_VAR` \u002F `_DATA_DIR_ENV_VAR`) redirect config + data — but the\n  CLI's **diagnostics logger ignores them**: it writes a per-invocation\n  `cli-*.log` under `state_dir()`, hardcoded to `Path.home()\u002F.omnigent\u002Flogs`\n  (`omnigent_ui_sdk\u002Fterminal\u002F_config.py`). So only redirecting `HOME` keeps a\n  non-help command (`config list`, the setup PTY spawns, `server stop`) from\n  writing into the real home. The driver does this for you.\n- `--strip-path` reduces `PATH` so `node`\u002F`tmux`\u002F`claude`\u002F`codex` read as \"not\n  installed\" → the true fresh-machine cold start.\n- Ambient model keys (`ANTHROPIC_API_KEY`, …) are stripped from the child env\n  unless you pass `--keep-env-creds`.\n\n**`--inherit-home` opts out** of `HOME` isolation — use it only to reach a real\ncredentialed REPL via ambient `~\u002F.claude` \u002F `~\u002F.databrickscfg` auth. It is\n**less safe**: a non-help command then writes `cli-*.log` into the real\n`~\u002F.omnigent\u002Flogs`.\n\nEvery run **fingerprints the real `~\u002F.omnigent` before and after** (stat-only,\nno content reads): the top-level config files **and** the set of\n`logs\u002Fcli-*.log` diagnostic files. A new config file\u002Fmtime *or* a new `cli-*.log`\nbasename trips `real_config_untouched: false`. With the default isolation that\nnever happens; under `--inherit-home` it correctly does — which is exactly the\nviolation the guard is meant to catch. If that check is ever `false`, stop and\ninvestigate. Run `check-isolation` first to confirm the loop is safe on your\nmachine.\n\n## Prerequisites\n\n- You're in the **worktree whose code you want to test** (each parallel agent\n  on its own worktree). The driver runs `omnigent` from `--repo`'s checkout.\n- A Python with `pexpect` — the project's `.venv\u002Fbin\u002Fpython` bundles it\n  (`pexpect>=4.9` in `pyproject.toml`). Run the driver with that interpreter.\n- An `omnigent` binary: the driver auto-finds `\u003Crepo>\u002F.venv\u002Fbin\u002Fomnigent`, or\n  pass `--omnigent \u003Cpath>`.\n- The setup \u002F picker \u002F help \u002F cold-start scenarios need **no credentials and no\n  harness**. Only `repl-commands` needs a working harness + credential: pass\n  `--inherit-home` (ambient `~\u002F.claude` auth) and\u002For `--keep-env-creds` (env API\n  key) with `--agent`. It reports `skipped`, never a false pass, when the prompt\n  isn't reachable.\n\n## Quick start\n\n```bash\nREPO=\u002Fpath\u002Fto\u002Fyour\u002Fworktree\nPY=$REPO\u002F.venv\u002Fbin\u002Fpython\nDRV=$REPO\u002F.claude\u002Fskills\u002Fcli-setup-verify\u002Fverify_cli.py\n\n# 0. Prove the sandbox is safe on this machine (do this once).\n#    HOME is isolated by default — no flag needed.\n$PY $DRV --scenario check-isolation --repo \"$REPO\"\n\n# 1. See exactly what a brand-new user sees on a fresh machine.\n$PY $DRV --scenario cold-start --strip-path --keep-sandbox --repo \"$REPO\"\n#    → reads the printed `artifacts` path, then `cat \u003Cthat>\u002Fcold_start.txt`\n\n# 2. Lint the top-level help (and any subcommand's).\n$PY $DRV --scenario help-snapshot --repo \"$REPO\"\n$PY $DRV --scenario help-snapshot --subcommand server --repo \"$REPO\"\n```\n\nEach run prints `SUMMARY {…}` and exits non-zero if any check failed (a\n`skipped` scenario exits 0). Pipe to `… | grep '^SUMMARY' | python -m json.tool`\nto read it.\n\n## Scenario catalog\n\n| Scenario | What it drives | Key checks \u002F notes | Maps to findings |\n|---|---|---|---|\n| `check-isolation` | `omnigent config list` in the sandbox (no PTY) | `config_list_ran`, `sandbox_config_home_used`, `real_config_untouched` | safety gate for everything |\n| `cold-start` | `omnigent setup` via PTY on a simulated fresh machine | `onboarding_rendered`, `harness_menu_present`; note `guided_default_affordance` | cold-start dead-end; missing \"recommended start here\" |\n| `setup-snapshot` | `omnigent setup`, optional `--nav-down N` arrow steps | `menu_rendered`; saves a frame per step | picker markers\u002Ffooter\u002Falignment; narrow-terminal at 80×24 |\n| `help-snapshot` | `omnigent [--subcommand] --help` (no PTY) | `help_rendered`, `no_param_leak`, `no_update_dup`; note `top_level_command_count` | `:param` leak, duplicate `update`\u002F`upgrade`, command sprawl |\n| `repl-commands` | `omnigent run \u003Cagent>` REPL, sends `\u002Fhelp` + `\u002Fquit` | `help_lists_commands`; note `quit_advertised` | REPL discoverability (`\u002Fhelp`, `\u002Fquit`) |\n\n`--list-scenarios` prints them too. Captured frames land in the printed\n`artifacts` dir as both `\u003Cname>.txt` (ANSI-stripped, for reading\u002Fasserting) and\n`\u003Cname>.ansi.txt` (raw, to see real colors with `less -R`).\n\n## The verifiable loop — a worked example\n\nFinding: *\"`server --help` leaks Sphinx `:param`\u002F`:returns` into user help.\"*\n\n```bash\n# BEFORE the fix (on the unfixed code):\n$PY $DRV --scenario help-snapshot --subcommand server --label before --repo \"$REPO\"\n#   → \"no_param_leak\": {\"ok\": false, ...}      ← bug reproduced (the baseline)\n\n# ... make the change (move :param docs into # comments) ...\n\n# AFTER the fix:\n$PY $DRV --scenario help-snapshot --subcommand server --label after --repo \"$REPO\"\n#   → \"no_param_leak\": {\"ok\": true, \"detail\": \"clean\"}   ← flipped → fix is verifiable\n```\n\nThe same shape proves the `update`\u002F`upgrade` duplicate (`no_update_dup`), the\ncold-start dead-end (`guided_default_affordance` note flips `absent`→`present`),\nor REPL `\u002Fquit` discoverability (`quit_advertised` note flips `no`→`yes`). **If\nthe check\u002Fnote doesn't flip, the fix isn't proven** — that is the signal to keep\nworking, and it's exactly the judgment the loop exists to force.\n\nIf a finding has no machine check yet, add one (see \"Adding a scenario\") so the\nfix becomes provable instead of asserted.\n\n## Examining UI\u002FUX deliberately\n\n- **Narrow terminal is the default.** The driver uses **80×24** — the size a\n  new user's window actually is, and where banner overflow and picker\n  redraw-past-the-bottom bugs appear. Re-run with `--cols 120 --rows 40` to\n  compare the roomy layout; diff the two frames.\n- **Read the frame, don't just trust the check.** `cat \u003Cartifacts>\u002Fcold_start.txt`\n  shows the literal screen — the all-`✗` menu, the footer hint (`Esc back` at\n  the root), the marker (`❯`), alignment of the status gutter. The frame *is*\n  the UX evidence.\n- **Compare pickers for consistency.** `setup-snapshot --nav-down 3` captures\n  the harness menu as you move; eyeball marker\u002Ffooter\u002Fhighlight drift against\n  the theme and resume pickers (different engines render differently).\n\n## Covering all critical user journeys\n\nThis skill owns the **setup \u002F onboarding \u002F first-run \u002F TUI** journeys. The repo\nalready has complementary CUJ coverage — use both:\n\n- **Live setup\u002FUX journeys → this skill's scenarios** (cold-start, setup,\n  pickers, help, REPL discoverability).\n- **Deeper end-to-end journeys → `tests\u002Fe2e\u002Ftest_journey_*.py`** (first session\n  to code, resume\u002Fdisconnect, fork\u002Fexplore, file upload, collaboration, …).\n  Run a slice with the project's gated runner, e.g.\n  `uv run --frozen --extra dev python -m pytest tests\u002Fe2e\u002Ftest_journey_first_session_to_code.py -q`.\n- **Reusable PTY helpers** live in `tests\u002Fe2e\u002Fomnigent\u002F_pexpect_harness.py`\n  (`spawn_omnigent_run`, `wait_for_ready`, `submit_prompt`, `await_turn_complete`,\n  `clean_exit`) and the snapshot comparator in `tests\u002Fe2e\u002Fomnigent\u002F_snapshot.py`\n  — prefer extending those over re-inventing.\n\nTo drive a surface this skill doesn't script yet, spawn it by hand with the\nsandbox env and the keys the driver exports (`KEY_UP`\u002F`KEY_DOWN`\u002F`KEY_ENTER`\u002F\n`KEY_ESC`), then `drain()` and `save_frame()` the result.\n\n## Teardown — non-negotiable\n\n- The driver force-kills the PTY child and its descendants, and runs\n  `omnigent server stop` against the sandbox to reap any spawned background\n  server. After a run, confirm nothing leaked:\n  `pgrep -af \"omnigent.*(server|runner|host._daemon)\"` — anything bound to your\n  sandbox's data dir is yours to kill.\n- The sandbox temp dir is deleted unless `--keep-sandbox`. If you keep one for\n  inspection, `rm -rf` it when done.\n- Always drive the CLI through the driver (which redirects `HOME` + the\n  config\u002Fdata knobs), never a bare `omnigent setup` — that would write to the\n  real `~\u002F.omnigent`. If you pass `--inherit-home`, expect `cli-*.log` writes to\n  the real `~\u002F.omnigent\u002Flogs` and a `real_config_untouched: false` — that's the\n  guard working, not a bug.\n\n## Honesty\n\nIf you can't reach the surface under test (no harness, no credential, headless\nlimit), the scenario must report `skipped` — **do not claim a CUJ passed**. The\nstrongest evidence for a fix is a reproduced baseline (`before`) plus the flipped\n`after`; report both `SUMMARY` lines, not a summary of a summary.\n\n## Adding a scenario\n\nWrite `scenario_\u003Cname>(args, sandbox, result)` in `verify_cli.py`: drive the CLI\n(reuse `pexpect.spawn(... env=sandbox.env, dimensions=(args.rows, args.cols))`,\n`drain()`, `save_frame()`, the `KEY_*` constants), record findings with\n`result.add(name, ok, detail)` (fails the run) or `result.notes.append(...)`\n(informational, for before\u002Fafter flips), register it in `SCENARIOS`, and add a\nrow to the catalog above. Keep one assertion per real, observable behavior so a\nfix is provable as a single check flip.\n\n## Code under test\n\n- First-run dispatch \u002F no-arg routing: `omnigent\u002Fcli.py` (`run`, the first-run\n  plan, `_run_configure_harnesses_interactive`).\n- Onboarding: `omnigent\u002Fonboarding\u002F*` (`setup.py`, `interactive.py`,\n  `configure_models.py`, `provider_selection.py`, `detected.py`).\n- TUI \u002F REPL & pickers: `omnigent\u002Frepl\u002F*` (`_repl.py`, `_theme_picker.py`,\n  `_resume_picker.py`), `omnigent\u002F_terminal_picker_theme.py`.\n- Installer: `scripts\u002Finstall_oss.sh`.\n","# Verifying the Omnigent CLI setup & UX in a closed loop\n\nThe Omnigent CLI's first impression is: `curl | sh` → run `omnigent` → pick a\nmodel credential → start a session. This skill lets an agent **enter that flow,\nexamine the UI\u002FUX, and prove whether a change is verifiable** — without a\nbrowser, without real credentials, and **without ever touching the developer's\nreal `~\u002F.omnigent`**.\n\nThe engine is `verify_cli.py` (next to this file). It drives the real\n`omnigent` binary through a pseudo-terminal (`pexpect`) inside a throwaway\nsandbox, captures what renders, runs assertions, and prints one machine-readable\n`SUMMARY {json}` line.\n\n> **The whole point is a verifiable loop**, not a one-shot check:\n> 1. Run a scenario on the **unfixed** code → baseline (`--label before`).\n> 2. Make the change.\n> 3. Run the **same** scenario → `--label after`.\n> 4. Diff the two `SUMMARY` lines. A fix is \"verifiable\" only if a concrete\n>    check or note **flips** between the two runs. If it doesn't flip, you\n>    can't prove the fix did anything — go back to step 2.\n\n## Why this is safe (read first)\n\nThe real `~\u002F.omnigent` here can be **many GB** (chat DB, runner logs, native\nharness state). The sandbox isolates every write three ways:\n\n- **`HOME` is redirected into the sandbox by default.** This is the load-bearing\n  one. `OMNIGENT_CONFIG_HOME` \u002F `OMNIGENT_DATA_DIR` (`omnigent\u002Fcli.py`\n  `_CONFIG_HOME_ENV_VAR` \u002F `_DATA_DIR_ENV_VAR`) redirect config + data — but the\n  CLI's **diagnostics logger ignores them**: it writes a per-invocation\n  `cli-*.log` under `state_dir()`, hardcoded to `Path.home()\u002F.omnigent\u002Flogs`\n  (`omnigent_ui_sdk\u002Fterminal\u002F_config.py`). So only redirecting `HOME` keeps a\n  non-help command (`config list`, the setup PTY spawns, `server stop`) from\n  writing into the real home. The driver does this for you.\n- `--strip-path` reduces `PATH` so `node`\u002F`tmux`\u002F`claude`\u002F`codex` read as \"not\n  installed\" → the true fresh-machine cold start.\n- Ambient model keys (`ANTHROPIC_API_KEY`, …) are stripped from the child env\n  unless you pass `--keep-env-creds`.\n\n**`--inherit-home` opts out** of `HOME` isolation — use it only to reach a real\ncredentialed REPL via ambient `~\u002F.claude` \u002F `~\u002F.databrickscfg` auth. It is\n**less safe**: a non-help command then writes `cli-*.log` into the real\n`~\u002F.omnigent\u002Flogs`.\n\nEvery run **fingerprints the real `~\u002F.omnigent` before and after** (stat-only,\nno content reads): the top-level config files **and** the set of\n`logs\u002Fcli-*.log` diagnostic files. A new config file\u002Fmtime *or* a new `cli-*.log`\nbasename trips `real_config_untouched: false`. With the default isolation that\nnever happens; under `--inherit-home` it correctly does — which is exactly the\nviolation the guard is meant to catch. If that check is ever `false`, stop and\ninvestigate. Run `check-isolation` first to confirm the loop is safe on your\nmachine.\n\n## Prerequisites\n\n- You're in the **worktree whose code you want to test** (each parallel agent\n  on its own worktree). The driver runs `omnigent` from `--repo`'s checkout.\n- A Python with `pexpect` — the project's `.venv\u002Fbin\u002Fpython` bundles it\n  (`pexpect>=4.9` in `pyproject.toml`). Run the driver with that interpreter.\n- An `omnigent` binary: the driver auto-finds `\u003Crepo>\u002F.venv\u002Fbin\u002Fomnigent`, or\n  pass `--omnigent \u003Cpath>`.\n- The setup \u002F picker \u002F help \u002F cold-start scenarios need **no credentials and no\n  harness**. Only `repl-commands` needs a working harness + credential: pass\n  `--inherit-home` (ambient `~\u002F.claude` auth) and\u002For `--keep-env-creds` (env API\n  key) with `--agent`. It reports `skipped`, never a false pass, when the prompt\n  isn't reachable.\n\n## Quick start\n\n```bash\nREPO=\u002Fpath\u002Fto\u002Fyour\u002Fworktree\nPY=$REPO\u002F.venv\u002Fbin\u002Fpython\nDRV=$REPO\u002F.claude\u002Fskills\u002Fcli-setup-verify\u002Fverify_cli.py\n\n# 0. Prove the sandbox is safe on this machine (do this once).\n#    HOME is isolated by default — no flag needed.\n$PY $DRV --scenario check-isolation --repo \"$REPO\"\n\n# 1. See exactly what a brand-new user sees on a fresh machine.\n$PY $DRV --scenario cold-start --strip-path --keep-sandbox --repo \"$REPO\"\n#    → reads the printed `artifacts` path, then `cat \u003Cthat>\u002Fcold_start.txt`\n\n# 2. Lint the top-level help (and any subcommand's).\n$PY $DRV --scenario help-snapshot --repo \"$REPO\"\n$PY $DRV --scenario help-snapshot --subcommand server --repo \"$REPO\"\n```\n\nEach run prints `SUMMARY {…}` and exits non-zero if any check failed (a\n`skipped` scenario exits 0). Pipe to `… | grep '^SUMMARY' | python -m json.tool`\nto read it.\n\n## Scenario catalog\n\n| Scenario | What it drives | Key checks \u002F notes | Maps to findings |\n|---|---|---|---|\n| `check-isolation` | `omnigent config list` in the sandbox (no PTY) | `config_list_ran`, `sandbox_config_home_used`, `real_config_untouched` | safety gate for everything |\n| `cold-start` | `omnigent setup` via PTY on a simulated fresh machine | `onboarding_rendered`, `harness_menu_present`; note `guided_default_affordance` | cold-start dead-end; missing \"recommended start here\" |\n| `setup-snapshot` | `omnigent setup`, optional `--nav-down N` arrow steps | `menu_rendered`; saves a frame per step | picker markers\u002Ffooter\u002Falignment; narrow-terminal at 80×24 |\n| `help-snapshot` | `omnigent [--subcommand] --help` (no PTY) | `help_rendered`, `no_param_leak`, `no_update_dup`; note `top_level_command_count` | `:param` leak, duplicate `update`\u002F`upgrade`, command sprawl |\n| `repl-commands` | `omnigent run \u003Cagent>` REPL, sends `\u002Fhelp` + `\u002Fquit` | `help_lists_commands`; note `quit_advertised` | REPL discoverability (`\u002Fhelp`, `\u002Fquit`) |\n\n`--list-scenarios` prints them too. Captured frames land in the printed\n`artifacts` dir as both `\u003Cname>.txt` (ANSI-stripped, for reading\u002Fasserting) and\n`\u003Cname>.ansi.txt` (raw, to see real colors with `less -R`).\n\n## The verifiable loop — a worked example\n\nFinding: *\"`server --help` leaks Sphinx `:param`\u002F`:returns` into user help.\"*\n\n```bash\n# BEFORE the fix (on the unfixed code):\n$PY $DRV --scenario help-snapshot --subcommand server --label before --repo \"$REPO\"\n#   → \"no_param_leak\": {\"ok\": false, ...}      ← bug reproduced (the baseline)\n\n# ... make the change (move :param docs into # comments) ...\n\n# AFTER the fix:\n$PY $DRV --scenario help-snapshot --subcommand server --label after --repo \"$REPO\"\n#   → \"no_param_leak\": {\"ok\": true, \"detail\": \"clean\"}   ← flipped → fix is verifiable\n```\n\nThe same shape proves the `update`\u002F`upgrade` duplicate (`no_update_dup`), the\ncold-start dead-end (`guided_default_affordance` note flips `absent`→`present`),\nor REPL `\u002Fquit` discoverability (`quit_advertised` note flips `no`→`yes`). **If\nthe check\u002Fnote doesn't flip, the fix isn't proven** — that is the signal to keep\nworking, and it's exactly the judgment the loop exists to force.\n\nIf a finding has no machine check yet, add one (see \"Adding a scenario\") so the\nfix becomes provable instead of asserted.\n\n## Examining UI\u002FUX deliberately\n\n- **Narrow terminal is the default.** The driver uses **80×24** — the size a\n  new user's window actually is, and where banner overflow and picker\n  redraw-past-the-bottom bugs appear. Re-run with `--cols 120 --rows 40` to\n  compare the roomy layout; diff the two frames.\n- **Read the frame, don't just trust the check.** `cat \u003Cartifacts>\u002Fcold_start.txt`\n  shows the literal screen — the all-`✗` menu, the footer hint (`Esc back` at\n  the root), the marker (`❯`), alignment of the status gutter. The frame *is*\n  the UX evidence.\n- **Compare pickers for consistency.** `setup-snapshot --nav-down 3` captures\n  the harness menu as you move; eyeball marker\u002Ffooter\u002Fhighlight drift against\n  the theme and resume pickers (different engines render differently).\n\n## Covering all critical user journeys\n\nThis skill owns the **setup \u002F onboarding \u002F first-run \u002F TUI** journeys. The repo\nalready has complementary CUJ coverage — use both:\n\n- **Live setup\u002FUX journeys → this skill's scenarios** (cold-start, setup,\n  pickers, help, REPL discoverability).\n- **Deeper end-to-end journeys → `tests\u002Fe2e\u002Ftest_journey_*.py`** (first session\n  to code, resume\u002Fdisconnect, fork\u002Fexplore, file upload, collaboration, …).\n  Run a slice with the project's gated runner, e.g.\n  `uv run --frozen --extra dev python -m pytest tests\u002Fe2e\u002Ftest_journey_first_session_to_code.py -q`.\n- **Reusable PTY helpers** live in `tests\u002Fe2e\u002Fomnigent\u002F_pexpect_harness.py`\n  (`spawn_omnigent_run`, `wait_for_ready`, `submit_prompt`, `await_turn_complete`,\n  `clean_exit`) and the snapshot comparator in `tests\u002Fe2e\u002Fomnigent\u002F_snapshot.py`\n  — prefer extending those over re-inventing.\n\nTo drive a surface this skill doesn't script yet, spawn it by hand with the\nsandbox env and the keys the driver exports (`KEY_UP`\u002F`KEY_DOWN`\u002F`KEY_ENTER`\u002F\n`KEY_ESC`), then `drain()` and `save_frame()` the result.\n\n## Teardown — non-negotiable\n\n- The driver force-kills the PTY child and its descendants, and runs\n  `omnigent server stop` against the sandbox to reap any spawned background\n  server. After a run, confirm nothing leaked:\n  `pgrep -af \"omnigent.*(server|runner|host._daemon)\"` — anything bound to your\n  sandbox's data dir is yours to kill.\n- The sandbox temp dir is deleted unless `--keep-sandbox`. If you keep one for\n  inspection, `rm -rf` it when done.\n- Always drive the CLI through the driver (which redirects `HOME` + the\n  config\u002Fdata knobs), never a bare `omnigent setup` — that would write to the\n  real `~\u002F.omnigent`. If you pass `--inherit-home`, expect `cli-*.log` writes to\n  the real `~\u002F.omnigent\u002Flogs` and a `real_config_untouched: false` — that's the\n  guard working, not a bug.\n\n## Honesty\n\nIf you can't reach the surface under test (no harness, no credential, headless\nlimit), the scenario must report `skipped` — **do not claim a CUJ passed**. The\nstrongest evidence for a fix is a reproduced baseline (`before`) plus the flipped\n`after`; report both `SUMMARY` lines, not a summary of a summary.\n\n## Adding a scenario\n\nWrite `scenario_\u003Cname>(args, sandbox, result)` in `verify_cli.py`: drive the CLI\n(reuse `pexpect.spawn(... env=sandbox.env, dimensions=(args.rows, args.cols))`,\n`drain()`, `save_frame()`, the `KEY_*` constants), record findings with\n`result.add(name, ok, detail)` (fails the run) or `result.notes.append(...)`\n(informational, for before\u002Fafter flips), register it in `SCENARIOS`, and add a\nrow to the catalog above. Keep one assertion per real, observable behavior so a\nfix is provable as a single check flip.\n\n## Code under test\n\n- First-run dispatch \u002F no-arg routing: `omnigent\u002Fcli.py` (`run`, the first-run\n  plan, `_run_configure_harnesses_interactive`).\n- Onboarding: `omnigent\u002Fonboarding\u002F*` (`setup.py`, `interactive.py`,\n  `configure_models.py`, `provider_selection.py`, `detected.py`).\n- TUI \u002F REPL & pickers: `omnigent\u002Frepl\u002F*` (`_repl.py`, `_theme_picker.py`,\n  `_resume_picker.py`), `omnigent\u002F_terminal_picker_theme.py`.\n- Installer: `scripts\u002Finstall_oss.sh`.\n",{"name":5,"description":7},"main","922625532f94df5ec745affe5f52a751e296a2e2",".claude\u002Fskills\u002Fcli-setup-verify","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel","2026-07-27 00:07:07",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add McClain-Thiel\u002FOmniSci --skill cli-setup-verify",{"tool":36,"command":37},"bunx","bunx skills add McClain-Thiel\u002FOmniSci --skill cli-setup-verify",{"tool":39,"command":40},"pnpm","pnpm dlx skills add McClain-Thiel\u002FOmniSci --skill cli-setup-verify","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcli-setup-verify\u002FSKILL.md",true,[44],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"verify_cli.py","SCRIPT","脚本",26744,"https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcli-setup-verify\u002Fverify_cli.py",[51,58,65,72,77,82],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":19},"01M79JBP2ZPFXG4NT9M82KX9Q0","deep-research","Research checklist and source-quality guidance for the archer e2e fixture.","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fexamples\u002Farcher\u002Fskills\u002Fdeep-research\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M79JBKPHZ40CQ5TM18JMFPKF","pi-e2e-xyz-greet-c4a8d5","Test greeter — pi skills e2e. The unique suffix c4a8d5 appears in the skill name so a string match in the agent's output proves Pi loaded this skill.","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_none\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[],{"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},"01M79JBG2WT8JHFR6R88TEQC8N","pi-e2e-xyz-count-d2f6e1","Test counter — pi skills e2e. The unique suffix d2f6e1 appears in the skill name so a string match in the agent's output proves Pi loaded this skill.","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_none\u002Fskills\u002Fpi-e2e-xyz-count-d2f6e1\u002FSKILL.md",[],[],{"skillNo":73,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M79JBC5A6FNJY4Q67K60K1PG","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_list\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[],{"skillNo":78,"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":79,"categories":80,"keywords":81,"publishedAt":19},"01M79JB8QHHYX07H6AWQ1BQ0NJ","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_list\u002Fskills\u002Fpi-e2e-xyz-count-d2f6e1\u002FSKILL.md",[],[],{"skillNo":83,"name":60,"nameZh":6,"description":61,"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},"01M79JB5SKWW13TENDM5MQWP0T","https:\u002F\u002Fgithub.com\u002FMcClain-Thiel\u002FOmniSci\u002Fblob\u002Fmain\u002Ftests\u002Fresources\u002Fagents\u002Fpi_skills_all\u002Fskills\u002Fpi-e2e-xyz-greet-c4a8d5\u002FSKILL.md",[],[]]