kolega-code 0.6.0__tar.gz → 0.7.1__tar.gz

{kolega_code-0.6.0 → kolega_code-0.7.1}/.env.example

@@ -3,6 +3,9 @@
 
 ANTHROPIC_API_KEY=
 OPENAI_API_KEY=
+# Prefer a ChatGPT subscription over an OpenAI API key? Run `/login` in the TUI
+# (no key needed). Optional override of the OAuth client id, for testing only:
+# KOLEGA_CODE_CHATGPT_CLIENT_ID=
 GOOGLE_API_KEY=
 MOONSHOT_API_KEY=
 DEEPSEEK_API_KEY=

{kolega_code-0.6.0 → kolega_code-0.7.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kolega-code
-Version: 0.6.0
+Version: 0.7.1
 Summary: Local-first AI coding agent for the terminal
 Project-URL: Homepage, https://kolega-ai.github.io/kolega-code/
 Project-URL: Documentation, https://kolega-ai.github.io/kolega-code/
@@ -43,9 +43,13 @@ Classifier: Topic :: Utilities
 Requires-Python: >=3.11
 Requires-Dist: anthropic==0.101.0
 Requires-Dist: beautifulsoup4==4.14.3
+Requires-Dist: ddgs==9.14.4
+Requires-Dist: firecrawl-py==4.28.2
 Requires-Dist: google-genai==2.0.1
+Requires-Dist: httpx==0.28.1
 Requires-Dist: jinja2==3.1.6
 Requires-Dist: langfuse==3.14.5
+Requires-Dist: lxml-html-clean==0.4.4
 Requires-Dist: nest-asyncio==1.6.0
 Requires-Dist: openai==1.109.1
 Requires-Dist: packaging==25.0
@@ -55,6 +59,7 @@ Requires-Dist: psutil==7.2.2
 Requires-Dist: pydantic==2.13.4
 Requires-Dist: python-dotenv==1.2.2
 Requires-Dist: pyyaml==6.0.3
+Requires-Dist: tavily-python==0.7.26
 Requires-Dist: tenacity==9.1.4
 Requires-Dist: textual==8.2.5
 Requires-Dist: tiktoken==0.12.0
@@ -103,8 +108,9 @@ sessions, settings, and API keys stay on your own machine.
   and applies precise edits.
 - **Runs commands.** Executes shell commands and watches their output in a dedicated terminal view.
 - **Plans before it builds.** Plan mode investigates and proposes; Build mode executes.
-- **Browses the web.** A built-in browser agent (powered by Playwright) fetches pages and
-  interacts with sites when a task needs it.
+- **Searches and browses the web.** The agent can search for relevant pages, fetch URLs
+  directly, or use the built-in browser agent (powered by Playwright) when a task needs
+  site interaction.
 - **Dispatches sub-agents.** Hands work off to specialized agents and tracks their activity live.
 - **Orchestrates workflows.** With [gigacode](https://kolega-ai.github.io/kolega-code/gigacode/),
   it fans out many sub-agents in parallel.
@@ -170,6 +176,14 @@ and DeepSeek — and lets you assign **different models to different roles**: a
 long-context model for coding, a fast cheap model for small utility calls, and one for extended
 "thinking". See [Providers & Models](https://kolega-ai.github.io/kolega-code/configuration/providers-and-models/).
 
+## Sign in with ChatGPT
+
+If you have a paid **ChatGPT** plan (Plus, Pro, or Business), you can use it to run OpenAI models
+instead of a separate API key. Run `/login chatgpt` in the TUI, complete the browser sign-in, and
+Kolega Code switches to the **OpenAI (ChatGPT subscription)** provider (default `gpt-5.5`). Tokens are
+stored locally (chmod `600`) and refreshed automatically; `/logout chatgpt` removes them. See
+[Sign in with ChatGPT](https://kolega-ai.github.io/kolega-code/configuration/sign-in-with-chatgpt/).
+
 ## Configuration
 
 Set your provider, model, and API keys from the **Settings** tab in the UI, or via environment
@@ -187,6 +201,10 @@ Settings. Local session state lives under your platform's state directory unless
 [Configuration docs](https://kolega-ai.github.io/kolega-code/configuration/settings-and-api-keys/)
 for the full story.
 
+The `web_search` tool uses DuckDuckGo by default without a key. To choose another backend,
+set it in **Settings** or export `KOLEGA_CODE_WEB_SEARCH_BACKEND` as `firecrawl`, `tavily`,
+or `searxng`; use `FIRECRAWL_API_KEY`, `TAVILY_API_KEY`, or `SEARXNG_BASE_URL` as needed.
+
 ## Requirements
 
 - **Python 3.11+**

{kolega_code-0.6.0 → kolega_code-0.7.1}/README.md

@@ -35,8 +35,9 @@ sessions, settings, and API keys stay on your own machine.
   and applies precise edits.
 - **Runs commands.** Executes shell commands and watches their output in a dedicated terminal view.
 - **Plans before it builds.** Plan mode investigates and proposes; Build mode executes.
-- **Browses the web.** A built-in browser agent (powered by Playwright) fetches pages and
-  interacts with sites when a task needs it.
+- **Searches and browses the web.** The agent can search for relevant pages, fetch URLs
+  directly, or use the built-in browser agent (powered by Playwright) when a task needs
+  site interaction.
 - **Dispatches sub-agents.** Hands work off to specialized agents and tracks their activity live.
 - **Orchestrates workflows.** With [gigacode](https://kolega-ai.github.io/kolega-code/gigacode/),
   it fans out many sub-agents in parallel.
@@ -102,6 +103,14 @@ and DeepSeek — and lets you assign **different models to different roles**: a
 long-context model for coding, a fast cheap model for small utility calls, and one for extended
 "thinking". See [Providers & Models](https://kolega-ai.github.io/kolega-code/configuration/providers-and-models/).
 
+## Sign in with ChatGPT
+
+If you have a paid **ChatGPT** plan (Plus, Pro, or Business), you can use it to run OpenAI models
+instead of a separate API key. Run `/login chatgpt` in the TUI, complete the browser sign-in, and
+Kolega Code switches to the **OpenAI (ChatGPT subscription)** provider (default `gpt-5.5`). Tokens are
+stored locally (chmod `600`) and refreshed automatically; `/logout chatgpt` removes them. See
+[Sign in with ChatGPT](https://kolega-ai.github.io/kolega-code/configuration/sign-in-with-chatgpt/).
+
 ## Configuration
 
 Set your provider, model, and API keys from the **Settings** tab in the UI, or via environment
@@ -119,6 +128,10 @@ Settings. Local session state lives under your platform's state directory unless
 [Configuration docs](https://kolega-ai.github.io/kolega-code/configuration/settings-and-api-keys/)
 for the full story.
 
+The `web_search` tool uses DuckDuckGo by default without a key. To choose another backend,
+set it in **Settings** or export `KOLEGA_CODE_WEB_SEARCH_BACKEND` as `firecrawl`, `tavily`,
+or `searxng`; use `FIRECRAWL_API_KEY`, `TAVILY_API_KEY`, or `SEARXNG_BASE_URL` as needed.
+
 ## Requirements
 
 - **Python 3.11+**

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/astro.config.mjs

@@ -56,6 +56,10 @@ export default defineConfig({
               label: "Providers & Models",
               slug: "configuration/providers-and-models",
             },
+            {
+              label: "Sign in with ChatGPT",
+              slug: "configuration/sign-in-with-chatgpt",
+            },
             {
               label: "Settings & API Keys",
               slug: "configuration/settings-and-api-keys",

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/concepts/tools.md

@@ -38,8 +38,14 @@ the [TUI](../../cli/overview/) or [`ask`](../../cli/ask/).
 
 ### Web
 
+- `web_search` — search the web for relevant pages and return ranked results.
 - `web_fetch` — fetch and parse a web page without a full browser.
 
+Use `web_search` when you do not already know the right URL, then follow up with
+`web_fetch` to read a result in depth. The default backend is keyless DuckDuckGo;
+Firecrawl, Tavily, and self-hosted SearXNG can be selected in Settings or with
+environment variables.
+
 ### Reasoning & memory
 
 - `think_hard` — an extended-reasoning step that uses the
@@ -57,8 +63,8 @@ and `dispatch_general_agent`.
 Tools are gated by mode. In a read-only context — like [Plan mode](../../tui/modes/)
 or an investigation sub-agent — only non-mutating tools are available
 (`list_directory`, `read_entire_file`, `read_file_section`, `search_codebase`,
-`find_files_by_pattern`, `web_fetch`, `think_hard`, and reading memory). Editing
-files and running commands require Build mode's full toolset.
+`find_files_by_pattern`, `web_search`, `web_fetch`, `think_hard`, and reading
+memory). Editing files and running commands require Build mode's full toolset.
 
 This separation is what makes Plan mode safe to run against any codebase: the
 planning agent can look but not touch.

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/configuration/environment-variables.md

@@ -87,6 +87,18 @@ section.
 | `KOLEGA_CODE_STATE_DIR` | Override where settings and sessions are stored |
 | `KOLEGA_CODE_ENVIRONMENT` | Environment label attached to tracing/metadata (default `development`) |
 
+## Web search
+
+The `web_search` tool defaults to the keyless DuckDuckGo backend. Set a backend
+explicitly when you want a cloud provider or a self-hosted SearXNG instance.
+
+| Variable | Purpose |
+| --- | --- |
+| `KOLEGA_CODE_WEB_SEARCH_BACKEND` | Backend for `web_search`: `duckduckgo`, `firecrawl`, `tavily`, or `searxng` |
+| `FIRECRAWL_API_KEY` | Optional Firecrawl key for higher rate limits |
+| `TAVILY_API_KEY` | Tavily API key |
+| `SEARXNG_BASE_URL` | Base URL for a self-hosted SearXNG instance |
+
 ## Telemetry (Langfuse)
 
 Optional [Langfuse](https://langfuse.com/) tracing of LLM usage.
@@ -113,6 +125,9 @@ KOLEGA_CODE_PROVIDER=moonshot
 KOLEGA_CODE_MODEL=kimi-k2.7-code
 KOLEGA_CODE_THINKING_EFFORT=auto
 
+# Optional: web search
+KOLEGA_CODE_WEB_SEARCH_BACKEND=duckduckgo
+
 # Optional: Langfuse tracing
 LANGFUSE_HOST=https://us.cloud.langfuse.com
 LANGFUSE_PUBLIC_KEY=

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/configuration/providers-and-models.mdx

@@ -16,6 +16,7 @@ The agent runtime recognizes these providers:
 | --- | --- | --- |
 | Anthropic | `anthropic` | `ANTHROPIC_API_KEY` |
 | OpenAI | `openai` | `OPENAI_API_KEY` |
+| OpenAI (ChatGPT subscription) | `openai_chatgpt` | _none — [sign in with ChatGPT](../sign-in-with-chatgpt/)_ |
 | Google | `google` | `GOOGLE_API_KEY` |
 | Groq | `groq` | `GROQ_API_KEY` |
 | Together.ai | `together` | `TOGETHER_API_KEY` |

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/configuration/settings-and-api-keys.mdx

@@ -32,6 +32,13 @@ agents (planning, building, investigation, general, browser) their own provider,
 model, and effort. Leave a role on **Default** to inherit the active model. See
 [Per-agent models](../providers-and-models/#per-agent-models).
 
+The **Web Search** section configures the `web_search` tool. DuckDuckGo is the
+default and needs no key. Firecrawl can also run without a key, with an optional
+key for higher rate limits. Tavily requires a key, and SearXNG requires the base
+URL of your self-hosted instance. The same settings can be supplied with
+[`KOLEGA_CODE_WEB_SEARCH_BACKEND`, `FIRECRAWL_API_KEY`, `TAVILY_API_KEY`, and
+`SEARXNG_BASE_URL`](../environment-variables/#web-search).
+
 ## Where settings live
 
 Settings are stored as JSON in your platform's state directory, in a file named
@@ -56,7 +63,8 @@ The file looks like this:
   "active_thinking_effort": "auto",
   "api_keys": {
     "moonshot": "sk-...",
-    "deepseek": "sk-..."
+    "deepseek": "sk-...",
+    "tavily": "tvly-..."
   },
   "agent_models": {
     "investigation": {
@@ -64,13 +72,17 @@ The file looks like this:
       "model": "deepseek-v4-flash",
       "thinking_effort": "high"
     }
-  }
+  },
+  "web_search_backend": "tavily",
+  "web_search_base_url": null
 }
 ```
 
 `agent_models` holds per-agent overrides keyed by role (`planning`, `building`,
 `investigation`, `general`, `browser`); a role that is absent inherits the active
 model. The field is optional and older `schema_version` files load unchanged.
+Cloud web-search keys are stored in `api_keys` under their backend names
+(`firecrawl` or `tavily`); `web_search_base_url` is used by SearXNG.
 
 <Aside type="caution" title="Keys are stored in plaintext">
 API keys are written to `settings.json` in plaintext, but the file is created with
@@ -83,7 +95,7 @@ store keys on disk at all, supply them via
 
 <FileTree>
 - kolega-code/
-  - settings.json   Provider, model, thinking effort, and API keys (0600)
+  - settings.json   Provider, model, web search, thinking effort, and API keys (0600)
   - sessions/
     - &lt;session-id&gt;.json   One file per saved session
 </FileTree>

kolega_code-0.7.1/docs/src/content/docs/configuration/sign-in-with-chatgpt.mdx

@@ -0,0 +1,93 @@
+---
+title: Sign in with ChatGPT
+description: Use your ChatGPT subscription to run OpenAI models in Kolega Code, without an API key.
+---
+
+import { Aside, Steps } from '@astrojs/starlight/components';
+
+If you have a paid **ChatGPT** plan, you can use it to run OpenAI models in Kolega
+Code — no separate `OPENAI_API_KEY` required. Sign in once with `/login chatgpt`
+and Kolega Code uses your subscription for the **OpenAI (ChatGPT subscription)**
+provider.
+
+## Eligibility
+
+| Plan | Runs models |
+| --- | --- |
+| Plus | Yes |
+| Pro | Yes |
+| Business | Yes |
+| Free | No |
+
+A Free plan can sign in, but can't run models this way — use an
+[`OPENAI_API_KEY`](../providers-and-models/) instead.
+
+## Signing in
+
+<Steps>
+
+1. In the TUI composer, run:
+
+   ```text
+   /login chatgpt
+   ```
+
+2. Your browser opens to the ChatGPT sign-in page. Complete the sign-in there.
+   (If the browser doesn't open, the composer prints a URL you can paste in
+   yourself.)
+
+3. When sign-in finishes, the browser shows a confirmation page and Kolega Code
+   switches to the **OpenAI (ChatGPT subscription)** provider, ready to use.
+
+</Steps>
+
+## Models
+
+The ChatGPT-subscription provider (identifier `openai_chatgpt`) exposes the Codex
+model line-up:
+
+| Model | Label |
+| --- | --- |
+| `gpt-5.5` | GPT-5.5 — default, frontier coding and reasoning |
+| `gpt-5.4` | GPT-5.4 — strong everyday coding |
+| `gpt-5.4-mini` | GPT-5.4 Mini — fast and economical |
+| `gpt-5.3-codex-spark` | GPT-5.3 Codex Spark — ultra-fast |
+
+Switch between them with [`/model`](../../tui/slash-commands/) or from the
+**Settings** tab. Each model supports a reasoning **thinking effort** you can set
+with [`/effort`](../../tui/slash-commands/).
+
+## Where credentials live
+
+Sign-in tokens are stored in your `settings.json` under `oauth_tokens`, in the
+same [state directory](../settings-and-api-keys/#where-settings-live) as the rest
+of your settings, with `0600` permissions (readable and writable only by your
+user). They refresh automatically in the background, so a single sign-in keeps
+working across sessions.
+
+## Checking status
+
+`doctor` reports whether you're signed in, without exposing any token:
+
+```bash
+kolega-code doctor --project .
+```
+
+For the ChatGPT-subscription provider it shows `signed in as <email> (<plan>)` or
+`not signed in`. See [`doctor`](../../cli/doctor/).
+
+## Signing out
+
+To remove your stored ChatGPT credentials:
+
+```text
+/logout chatgpt
+```
+
+After signing out, switch to another provider — or sign in again with
+`/login chatgpt` — to keep working.
+
+<Aside type="tip">
+`/login` and `/logout` take a provider name so more sign-in options can be added
+over time. Today, `chatgpt` is the available target.
+</Aside>

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/index.mdx

@@ -19,8 +19,9 @@ and your sessions, settings, and API keys are stored on your own machine.
   dedicated terminal view in the UI.
 - **Plans before it builds.** A read-only **Plan mode** produces a reviewable plan
   and a shared task list; **Build mode** implements it.
-- **Browses the web.** A built-in browser agent (powered by Playwright) can fetch
-  pages and interact with sites when a task needs it.
+- **Searches and browses the web.** The agent can search for relevant pages,
+  fetch URLs directly, or use the built-in browser agent (powered by Playwright)
+  when a task needs site interaction.
 - **Dispatches sub-agents.** For larger tasks, the main agent can hand off to
   specialized agents (investigation, browser, coding, general) and track their
   activity live.

{kolega_code-0.6.0 → kolega_code-0.7.1}/docs/src/content/docs/tui/slash-commands.md

@@ -40,6 +40,8 @@ These control the app and your session.
 | `/permissions` | Show or switch the shell/edit permission mode |
 | `/model` | Choose the active model |
 | `/effort` | Choose the active model's thinking effort |
+| `/login` | Sign in to a provider, e.g. `/login chatgpt` |
+| `/logout` | Sign out of a provider, e.g. `/logout chatgpt` |
 | `/gigacode` | Toggle [gigacode](../../gigacode/) workflow orchestration on or off |
 | `/copy` | Copy the last response to the clipboard |
 | `/version` | Show the Kolega Code version |
@@ -53,6 +55,10 @@ provider. You can also switch directly with `/model <name>`.
 Run `/effort` to open a selectable list of supported effort values for the
 active model. You can also switch directly with `/effort <level>`.
 
+Run `/login chatgpt` to sign in with a ChatGPT subscription and use OpenAI models
+without an API key; `/logout chatgpt` removes the stored credentials. See
+[Sign in with ChatGPT](../../configuration/sign-in-with-chatgpt/).
+
 Run `/init` to have the agent inspect the repository and create or update a
 concise root `AGENTS.md`. Extra text after the command is passed as focus or
 constraints:

{kolega_code-0.6.0 → kolega_code-0.7.1}/kolega_code/__init__.py

@@ -148,4 +148,4 @@ __all__ = [
     "parse_git_status_output",
 ]
 
-__version__ = "0.6.0"
+__version__ = "0.7.1"

{kolega_code-0.6.0 → kolega_code-0.7.1}/kolega_code/agent/__init__.py

@@ -39,4 +39,4 @@ __all__ = [
     "ToolExtension",
 ]
 
-__version__ = "0.6.0"
+__version__ = "0.7.1"

{kolega_code-0.6.0 → kolega_code-0.7.1}/kolega_code/agent/baseagent.py

@@ -936,6 +936,7 @@ class BaseAgent(LogMixin):
             max_retries=model_config.rate_limits.max_retries,
             requests_per_minute=model_config.rate_limits.requests_per_minute,
             tokens_per_minute=model_config.rate_limits.tokens_per_minute,
+            token_manager=self.config.get_chatgpt_token_manager(),
         )
         response = await client.generate(
             model=model_config.model,

{kolega_code-0.6.0 → kolega_code-0.7.1}/kolega_code/agent/context.py

@@ -113,6 +113,7 @@ class AgentContext:
                 user_id=self.telemetry.user_id,
                 user_email=self.telemetry.user_email,
                 usage_recorder=self.telemetry.usage_recorder,
+                token_manager=self.config.get_chatgpt_token_manager(),
             )
 
         return LLMClient(
@@ -121,4 +122,5 @@ class AgentContext:
             max_retries=model_config.rate_limits.max_retries,
             requests_per_minute=model_config.rate_limits.requests_per_minute,
             tokens_per_minute=model_config.rate_limits.tokens_per_minute,
+            token_manager=self.config.get_chatgpt_token_manager(),
         )

kolega_code-0.7.1/kolega_code/agent/tests/llm/test_chatgpt_live.py

@@ -0,0 +1,101 @@
+"""Live test for the ChatGPT-subscription (OAuth) provider.
+
+Unlike the key-based providers, this one needs OAuth tokens from an interactive
+``/login``. It is skipped unless real tokens are available, so it never runs in
+CI. To run it locally after signing in via the TUI::
+
+    pytest -m integration kolega_code/agent/tests/llm/test_chatgpt_live.py -v
+
+Token source (first match wins):
+  - KOLEGA_CODE_CHATGPT_TOKENS env var holding the JSON of an OAuthTokens dump, or
+  - the stored token in the local settings file (the one /login writes).
+"""
+
+import json
+import os
+
+import pytest
+
+from kolega_code.auth.tokens import ChatGPTTokenManager, OAuthTokens
+from kolega_code.auth import constants as chatgpt_constants
+from kolega_code.cli.settings import SettingsStore
+from kolega_code.llm.client import LLMClient
+from kolega_code.llm.models import Message, MessageHistory, TextBlock
+
+pytestmark = pytest.mark.integration
+
+SKIP_IN_CI = bool(os.getenv("CI")) or bool(os.getenv("GITLAB_CI"))
+
+
+def _load_tokens() -> OAuthTokens | None:
+    raw = os.getenv("KOLEGA_CODE_CHATGPT_TOKENS")
+    if raw:
+        try:
+            return OAuthTokens.model_validate(json.loads(raw))
+        except (ValueError, TypeError):
+            return None
+    stored = SettingsStore().load().get_oauth_token(chatgpt_constants.PROVIDER_KEY)
+    if stored:
+        try:
+            return OAuthTokens.model_validate(stored)
+        except (ValueError, TypeError):
+            return None
+    return None
+
+
+def _require_tokens() -> OAuthTokens:
+    if SKIP_IN_CI:
+        pytest.skip("Skipping live ChatGPT call in CI")
+    tokens = _load_tokens()
+    if tokens is None:
+        pytest.skip("No ChatGPT tokens available; sign in with /login or set KOLEGA_CODE_CHATGPT_TOKENS")
+    return tokens
+
+
+@pytest.mark.asyncio
+async def test_live_chatgpt_generate() -> None:
+    tokens = _require_tokens()
+    client = LLMClient(
+        provider=chatgpt_constants.PROVIDER_KEY,
+        api_key="unused",
+        token_manager=ChatGPTTokenManager(tokens),
+    )
+
+    response = await client.generate(
+        messages=MessageHistory(
+            [Message(role="user", content=[TextBlock(text="What is 2 + 2? Reply with just the number.")])]
+        ),
+        system=Message(role="system", content=[TextBlock(text="You are concise.")]),
+        model=chatgpt_constants.DEFAULT_MODEL,
+        max_completion_tokens=4096,
+        thinking="low",
+    )
+
+    assert response.role == "assistant"
+    assert response.get_text_content(), "empty response from ChatGPT backend"
+
+
+@pytest.mark.asyncio
+async def test_live_chatgpt_stream() -> None:
+    tokens = _require_tokens()
+    client = LLMClient(
+        provider=chatgpt_constants.PROVIDER_KEY,
+        api_key="unused",
+        token_manager=ChatGPTTokenManager(tokens),
+    )
+
+    text = ""
+    async with await client.stream(
+        messages=MessageHistory(
+            [Message(role="user", content=[TextBlock(text="Say the word 'ready' and nothing else.")])]
+        ),
+        model=chatgpt_constants.DEFAULT_MODEL,
+        max_completion_tokens=4096,
+        thinking="low",
+    ) as stream:
+        async for chunk in stream:
+            if chunk.type == "text" and chunk.text:
+                text += chunk.text
+        message = await stream.get_final_message()
+
+    assert text or message.get_text_content(), "empty stream from ChatGPT backend"