litesquad 0.0.1__tar.gz

litesquad-0.0.1/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

litesquad-0.0.1/PKG-INFO

@@ -0,0 +1,55 @@
+Metadata-Version: 2.4
+Name: litesquad
+Version: 0.0.1
+Summary: Lightweight tool for working with a team of LLMs
+Author: EricThomson
+Author-email: EricThomson <thomson.eric@gmail.com>
+License-Expression: Apache-2.0
+License-File: LICENSE
+Requires-Dist: litellm>=1.90.1
+Requires-Dist: pydantic>=2.13.4
+Requires-Dist: python-dotenv>=1.0.0
+Requires-Dist: rich>=15.0.0
+Requires-Dist: typer>=0.26.8
+Requires-Dist: dash>=3.0,<4.2 ; extra == 'web'
+Requires-Python: >=3.12
+Project-URL: Repository, https://github.com/EricThomson/litesquad
+Provides-Extra: web
+Description-Content-Type: text/markdown
+
+# lite squad
+Tool for working with a team of LLMs. Sometimes two, or five, heads are better than one.
+
+## Usage
+Install with pip:
+
+    pip install litesquad
+
+Use in browser: 
+
+    litesquad --web 
+
+Use at command line:
+
+    litesquad "plan such and such project"
+    litesquad "such and such" --quick # get a quick answer, bypassing the squad
+
+litesquad takes a few minutes to run when not in deep think mode. There's a lot of LLM calls going on under the hood (see below). 
+
+Query is distributed to worker LLMs (gemini and sonnet direct, plus deepseek, mistral, and llama via OpenRouter). Another (grok) acts as a critic that gives feedback to the workers. They revise their response. An intermediate representation of these responses is extracted to pull the content and clustered into categories of suggestions (gpt5). A judge (opus) converts these suggestions into a final coherent answer for the user. 
+
+To run the default, you will need API keys for [Gemini](https://aistudio.google.com/app/apikey), [OpenAI](https://openai.com/index/openai-api/), [Anthropic](https://platform.claude.com/docs/en/api/admin/api_keys/retrieve), and [OpenRouter](https://openrouter.ai/keys). One OpenRouter key reaches the whole openrouter.ai catalog (deepseek, mistral, llama, grok, qwen, ...), which is how the worker roster grows wide without a key per provider. You can store your API keys in `.env`.
+
+### Basic tests
+To see if api keys are working (from activated env):  `python -m litesquad.check_keys`
+
+Check on specific models: `litesquad --check` 
+
+Offline test: `litesquad --smoke --mock`
+
+
+### Caveats
+This provides an interface to an ensemble of LLMs to try to generate a better answer than when using a single LLM. No agentic tool-usage, such as web calls, from the LLMs. Just vanilla text processing.
+
+With apologies to [squad](https://github.com/bradygaster/squad). 
+

litesquad-0.0.1/README.md

@@ -0,0 +1,36 @@
+# lite squad
+Tool for working with a team of LLMs. Sometimes two, or five, heads are better than one.
+
+## Usage
+Install with pip:
+
+    pip install litesquad
+
+Use in browser: 
+
+    litesquad --web 
+
+Use at command line:
+
+    litesquad "plan such and such project"
+    litesquad "such and such" --quick # get a quick answer, bypassing the squad
+
+litesquad takes a few minutes to run when not in deep think mode. There's a lot of LLM calls going on under the hood (see below). 
+
+Query is distributed to worker LLMs (gemini and sonnet direct, plus deepseek, mistral, and llama via OpenRouter). Another (grok) acts as a critic that gives feedback to the workers. They revise their response. An intermediate representation of these responses is extracted to pull the content and clustered into categories of suggestions (gpt5). A judge (opus) converts these suggestions into a final coherent answer for the user. 
+
+To run the default, you will need API keys for [Gemini](https://aistudio.google.com/app/apikey), [OpenAI](https://openai.com/index/openai-api/), [Anthropic](https://platform.claude.com/docs/en/api/admin/api_keys/retrieve), and [OpenRouter](https://openrouter.ai/keys). One OpenRouter key reaches the whole openrouter.ai catalog (deepseek, mistral, llama, grok, qwen, ...), which is how the worker roster grows wide without a key per provider. You can store your API keys in `.env`.
+
+### Basic tests
+To see if api keys are working (from activated env):  `python -m litesquad.check_keys`
+
+Check on specific models: `litesquad --check` 
+
+Offline test: `litesquad --smoke --mock`
+
+
+### Caveats
+This provides an interface to an ensemble of LLMs to try to generate a better answer than when using a single LLM. No agentic tool-usage, such as web calls, from the LLMs. Just vanilla text processing.
+
+With apologies to [squad](https://github.com/bradygaster/squad). 
+

litesquad-0.0.1/pyproject.toml

@@ -0,0 +1,47 @@
+[build-system]
+requires = ["uv_build>=0.11.26,<0.12.0"]
+build-backend = "uv_build"
+
+[project]
+name = "litesquad"
+version = "0.0.1"
+description = "Lightweight tool for working with a team of LLMs"
+license = "Apache-2.0"
+license-files = ["LICENSE"]
+readme = "README.md"
+authors = [
+    { name = "EricThomson", email = "thomson.eric@gmail.com" }
+]
+requires-python = ">=3.12"
+dependencies = [
+    "litellm>=1.90.1",
+    "pydantic>=2.13.4",
+    "python-dotenv>=1.0.0",
+    "rich>=15.0.0",
+    "typer>=0.26.8",
+]
+
+[project.optional-dependencies]
+web = [
+    # <4.2: Dash 4.2+ remounts every child whenever a callback writes a children
+    # prop (open regression, plotly/dash#3846), which wipes DOM state like
+    # <details> open/closed in the live view. Lift the cap when that issue closes.
+    "dash>=3.0,<4.2",
+]
+
+[project.scripts]
+litesquad = "litesquad:main"
+litesquad-keys = "litesquad.check_keys:main"
+
+[dependency-groups]
+dev = [
+    "pytest>=9.1.1",
+    "ruff>=0.15.20",
+]
+
+[project.urls]
+Repository = "https://github.com/EricThomson/litesquad"
+
+[tool.uv.build-backend]
+module-root = "src"
+

litesquad-0.0.1/src/litesquad/__init__.py

@@ -0,0 +1,5 @@
+from .cli import app
+
+
+def main() -> None:
+    app()

litesquad-0.0.1/src/litesquad/check_keys.py

@@ -0,0 +1,57 @@
+"""Dead-simple check that each provider API key is valid and active.
+
+Run:  uv run litesquad-keys      (or: uv run python -m litesquad.check_keys)
+
+Pings one cheap model per provider with a 1-token request and prints a
+checkmark per key. Independent of the squad config, so it tests the key
+itself, not whichever model your squad happens to use.
+"""
+
+import os
+
+import litellm
+from rich.console import Console
+
+from .llm import load_env  # loads .env with override=True; importing also quiets litellm
+
+console = Console()
+
+# One cheap, broadly-available probe model per provider key.
+PROBES = {
+    "ANTHROPIC_API_KEY": "anthropic/claude-haiku-4-5",
+    "OPENAI_API_KEY": "openai/gpt-4o-mini",
+    "GEMINI_API_KEY": "gemini/gemini-2.5-flash",
+    "OPENROUTER_API_KEY": "openrouter/openai/gpt-4o-mini",
+}
+
+
+def check(model: str) -> tuple[bool, str]:
+    try:
+        litellm.completion(
+            model=model,
+            messages=[{"role": "user", "content": "ping"}],
+            max_tokens=1,
+            drop_params=True,
+        )
+        return True, "ok"
+    except Exception as exc:  # noqa: BLE001 - any failure means the key isn't usable
+        return False, str(exc).splitlines()[0].strip()
+
+
+def main() -> None:
+    load_env()
+    for key, model in PROBES.items():
+        # An unset key is a distinct state from an invalid one: say "not set"
+        # instead of probing and printing a provider auth exception.
+        if not os.environ.get(key):
+            console.print(f"[red]✗[/] {key} — not set")
+            continue
+        ok, detail = check(model)
+        if ok:
+            console.print(f"[green]✓[/] {key} — active")
+        else:
+            console.print(f"[red]✗[/] {key} — invalid: {detail}")
+
+
+if __name__ == "__main__":
+    main()

litesquad-0.0.1/src/litesquad/cli.py

@@ -0,0 +1,276 @@
+"""Command-line interface: one fixed-flow run, then interactive follow-ups."""
+
+from datetime import datetime
+from pathlib import Path
+
+import typer
+from rich.console import Console
+from rich.markdown import Markdown
+from rich.panel import Panel
+from rich.prompt import Prompt
+from rich.status import Status
+
+from . import paths
+from .config import RunConfig, SquadConfig, ensure_starter, load_config
+from .llm import LLMError, MissingKeysError, call_model, load_env, mock_call_model, preflight
+from .models import Conversation, Stage, TranscriptEvent
+from .squad import run_quick, run_turn
+
+app = typer.Typer(add_completion=False, help="Ask a small ensemble of LLMs anything.")
+console = Console()
+
+STAGE_LABEL: dict[Stage, str] = {
+    "propose": "responding",
+    "critique": "critiquing",
+    "revise": "revising",
+    "extract": "de-stylizing",
+    "cluster": "clustering",
+    "judge": "judging",
+    "reply": "answering",
+}
+QUIT_WORDS = {":quit", ":q", "quit", "exit"}
+SMOKE_PROMPT = "What is 1 + 1? Answer in one short sentence."
+
+
+class ConsoleReporter:
+    """Renders each stage as a Rich panel and appends events to a JSONL file.
+
+    Worker chains run in parallel, so their stages complete interleaved. Printing them
+    as they land is unreadable, so chain stages (propose/critique/revise) are buffered
+    per worker and printed as one coherent block the moment that worker's chain finishes
+    (or dies) -- blocks appear in chain-completion order. Everything else (extract,
+    cluster, judge, reply) prints as it completes, and the spinner always shows everyone
+    currently in flight. The transcript JSONL is still appended per event in completion
+    order: it is the durable stream, and the web UI regroups it by worker anyway.
+
+    Per the Reporter contract, calls arrive serialized, so no lock is needed here;
+    printing while the Status runs is safe (rich renders it above the spinner).
+    """
+
+    def __init__(self, transcript_path: Path | None) -> None:
+        self.transcript_path = transcript_path
+        self._status: Status | None = None
+        self._in_flight: dict[tuple[str, Stage], str] = {}  # (role, stage) -> spinner text
+        self._pending_chains: dict[str, list[Panel]] = {}  # worker key -> panels so far
+
+    @staticmethod
+    def _worker_key(role: str) -> str:
+        """propose/revise carry ``worker_N``; critique carries ``critic->worker_N``."""
+        return role.split("->")[-1]
+
+    @staticmethod
+    def _panel(event: TranscriptEvent) -> Panel:
+        title = f"{event.role} | {event.model}"
+        if event.error:
+            return Panel(event.error, title=f"{title} (error)", border_style="red")
+        border = "green" if event.stage in ("judge", "reply") else "cyan"
+        return Panel(Markdown(event.output), title=title, border_style=border)
+
+    def _spinner_text(self) -> str:
+        return "  |  ".join(self._in_flight.values())
+
+    def _flush_chain(self, worker: str) -> None:
+        for panel in self._pending_chains.pop(worker, []):
+            console.print(panel)
+
+    def stage_start(self, stage: Stage, role: str, model: str) -> None:
+        self._in_flight[(role, stage)] = f"[bold]{role}[/] ({model}) - {STAGE_LABEL[stage]}..."
+        if self._status is None:
+            self._status = console.status(self._spinner_text(), spinner="dots")
+            self._status.start()
+        else:
+            self._status.update(self._spinner_text())
+
+    def stage_done(self, event: TranscriptEvent) -> None:
+        self._in_flight.pop((event.role, event.stage), None)
+        if self.transcript_path is not None:
+            with self.transcript_path.open("a", encoding="utf-8") as fh:
+                fh.write(event.to_jsonl() + "\n")
+        if event.stage in ("propose", "critique", "revise"):
+            worker = self._worker_key(event.role)
+            self._pending_chains.setdefault(worker, []).append(self._panel(event))
+            if event.stage == "revise" or event.error:  # chain finished, or died here
+                self._flush_chain(worker)
+        else:
+            console.print(self._panel(event))
+        if self._status is not None:
+            if self._in_flight:
+                self._status.update(self._spinner_text())
+            else:
+                self._status.stop()
+                self._status = None
+
+    def close(self) -> None:
+        """Flush any unfinished chains and stop the spinner. Call in a ``finally``: an
+        aborted turn must still show what its surviving chains produced, and must not
+        leave a live display running (hidden cursor, endless spinner)."""
+        for worker in list(self._pending_chains):
+            self._flush_chain(worker)
+        if self._status is not None:
+            self._status.stop()
+            self._status = None
+        self._in_flight.clear()
+
+
+def _transcript_path(save: bool) -> Path | None:
+    if not save:
+        return None
+    stamp = datetime.now().strftime("%Y-%m-%d_%H%M%S")
+    return paths.transcripts_dir() / f"{stamp}.jsonl"
+
+
+def _model_roles(config: SquadConfig) -> list[tuple[str, str]]:
+    """Distinct configured models in pipeline order (workers first, then critic ->
+    extractor -> clusterer -> judge), each with the role(s) it holds -- so --check
+    doubles as a who-does-what roster card."""
+    assignments = [
+        *(("worker", worker.model) for worker in config.workers),
+        ("critic", config.critic.model),
+        ("extractor", config.extractor.model),
+        ("clusterer", config.clusterer.model),
+        ("judge", config.judge.model),
+    ]
+    roles_by_model: dict[str, list[str]] = {}
+    for role, model in assignments:
+        roles = roles_by_model.setdefault(model, [])
+        if role not in roles:  # two workers on one model still read as one "worker"
+            roles.append(role)
+    return [(model, ", ".join(roles)) for model, roles in roles_by_model.items()]
+
+
+def _check_models(config: SquadConfig) -> bool:
+    """Ping each distinct configured model with a tiny request. Returns all-ok.
+
+    ``max_tokens`` is generous (not 5) because reasoning models (GPT-5,
+    Gemini 2.5 Pro) spend output budget on hidden reasoning before any visible
+    text — too small a cap returns empty content. The cap only prevents
+    truncation; actual usage stays tiny since the model stops after "ok".
+    """
+    run_cfg = RunConfig(max_tokens=1024, save_transcript=False)
+    messages = [{"role": "user", "content": "Reply with the single word: ok"}]
+    all_ok = True
+    for model, roles in _model_roles(config):
+        try:
+            reply = call_model(model, messages, run_cfg, role="check").strip()
+            console.print(f"[green]✓[/] {model} [dim]({roles})[/] — {reply[:40]}")
+        except LLMError as exc:
+            all_ok = False
+            console.print(f"[red]✗[/] {model} [dim]({roles})[/] — {exc}")
+    return all_ok
+
+
+@app.command()
+def run(
+    task: str = typer.Argument(None, help="Your question or task for the ensemble."),
+    quick: bool = typer.Option(
+        False, "--quick", help="Talk to just the judge (Opus), skipping the ensemble."
+    ),
+    mock: bool = typer.Option(
+        False, "--mock", help="Use canned offline responses; no API keys needed."
+    ),
+    check: bool = typer.Option(
+        False, "--check", help="Ping each configured model with a tiny request and exit."
+    ),
+    smoke: bool = typer.Option(
+        False, "--smoke", help="Run one real turn on a fixed tiny prompt and exit (cheap end-to-end)."
+    ),
+    web: bool = typer.Option(
+        False, "--web", help="Serve the web UI instead of the terminal (needs the 'web' extra)."
+    ),
+    port: int = typer.Option(8050, help="Port for the web UI (only used with --web)."),
+) -> None:
+    """Ask the ensemble (or, with --quick, just the judge), then take follow-ups."""
+    if web:
+        if task is not None:
+            console.print("[red]With --web, submit tasks in the browser, not on the command line.[/]")
+            raise typer.Exit(2)
+        from .web import serve  # lazy: dash is an optional extra
+
+        serve(port=port, mock=mock)
+        return
+
+    load_env()
+
+    cfg_path = paths.config_path()
+    if ensure_starter(cfg_path):
+        console.print(
+            f"[dim]Wrote a starter config (all defaults, commented) you can edit at {cfg_path}[/]"
+        )
+    config = load_config(cfg_path)
+
+    if check:
+        try:
+            preflight(config)
+        except MissingKeysError as exc:
+            console.print(f"[red]{exc}[/]")
+            raise typer.Exit(1) from exc
+        console.print("Pinging configured models…")
+        raise typer.Exit(0 if _check_models(config) else 1)
+
+    if smoke:
+        caller = mock_call_model if mock else call_model
+        if not mock:
+            try:
+                preflight(config)
+            except MissingKeysError as exc:
+                console.print(f"[red]{exc}[/]")
+                raise typer.Exit(1) from exc
+        console.print(f'Smoke test: one turn on [dim]"{SMOKE_PROMPT}"[/]\n')
+        transcript_path = _transcript_path(config.run.save_transcript)
+        reporter = ConsoleReporter(transcript_path)
+        try:
+            run_turn(Conversation(), SMOKE_PROMPT, config, reporter, caller=caller)
+        except Exception as exc:  # noqa: BLE001 - smoke surfaces ANY failure (call or save)
+            console.print(f"[red]Smoke test FAILED: {exc}[/]")
+            raise typer.Exit(1) from exc
+        finally:
+            reporter.close()
+        if transcript_path is not None:
+            console.print(f"[dim]Transcript: {transcript_path}[/]")
+        console.print("[green]Smoke test passed - all stages produced output.[/]")
+        raise typer.Exit(0)
+
+    if task is None:
+        console.print('[red]Provide a task, e.g. litesquad "Plan my week", or use --check.[/]')
+        raise typer.Exit(2)
+
+    caller = call_model
+    if mock:
+        caller = mock_call_model
+        console.print("[yellow]Running in --mock mode: canned responses, no API calls.[/]")
+    else:
+        try:
+            preflight(config)
+        except MissingKeysError as exc:
+            console.print(f"[red]{exc}[/]")
+            raise typer.Exit(1) from exc
+
+    conversation = Conversation()
+    transcript_path = _transcript_path(config.run.save_transcript)
+    reporter = ConsoleReporter(transcript_path)
+    run_one = run_quick if quick else run_turn
+    if quick:
+        console.print("[dim]Quick mode: just the judge, no ensemble.[/]")
+
+    current_task = task
+    while True:
+        try:
+            run_one(conversation, current_task, config, reporter, caller=caller)
+        except LLMError as exc:
+            console.print(f"[red]Turn aborted: {exc}[/]")
+        finally:
+            reporter.close()
+
+        if transcript_path is not None:
+            console.print(f"[dim]Transcript: {transcript_path}[/]")
+
+        try:
+            reply = Prompt.ask(
+                "\n[bold magenta]Follow-up[/] ([bold yellow]:quit to exit[/])"
+            ).strip()
+        except (EOFError, KeyboardInterrupt):
+            console.print()
+            break
+        if not reply or reply.lower() in QUIT_WORDS:
+            break
+        current_task = reply