{"slug": "show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli", "title": "Show HN: Frontier model pricing became a rip-off, so I built an open-source CLI", "summary": "Kolega Code, an open-source CLI tool for multi-agent coding in the terminal, has been released by developer Kolega AI. The tool features Gigacode orchestration that fans out specialized sub-agents for parallel work, supports model routing, web search, browser automation, and MCP servers, and keeps all state local. It is available via curl or uv/pip install.", "body_md": "**Multi-agent coding in the terminal.**\n\nKolega Code is a local-first terminal coding agent built for work that is too wide\nfor one loop: fan out specialized sub-agents with **Gigacode**, route different\nmodels to different jobs, search the web, drive a browser, and keep sessions,\nsettings, permissions, and credentials on your machine.\n\nMost terminal agents are strongest when one model can reason through one task at a time. Kolega Code keeps that familiar workflow, then adds orchestration for broad work: large audits, sweeping migrations, cross-file checks, adversarial reviews, and implementation plans whose pieces can run independently.\n\nWith [Gigacode](https://kolega-ai.github.io/kolega-code/gigacode/), Kolega Code can:\n\n**Fan out many sub-agents at once.** Split a wide codebase review by package, assign independent implementation tasks, or run checks across many directories in parallel.**Use real workflow shapes.** The agent can generate parallel phases, pipelines, loops, judge panels, and synthesis steps instead of only delegating one task at a time.**Keep orchestration visible.** Workflow phase headers and progress lines appear in the transcript; the sub-agent inspector shows each agent's live trajectory.**Save inspectable artifacts.** Each run keeps the generated workflow script, result files, a Markdown transcript, raw JSONL, a resume journal, and debug sub-agent transcripts under Kolega Code's state directory.**Run in either mode.** In Plan mode, workflow sub-agents stay read-only for parallel investigation. In Build mode, they can use the full coding toolset.**Resume interrupted runs.** Finished workflow steps are journaled so a deliberate resume does not have to restart the whole fan-out.\n\nUse normal chat for focused changes. Turn on Gigacode when the problem is broad enough that one serial agent pass would be the bottleneck.\n\n**Gigacode orchestration:** parallel, pipelined, looped, judged, and synthesized multi-agent workflows with saved artifacts and resume support.**Specialized sub-agents:** planning, building/coder, investigation, general, and browser agents, with live activity tracking in the TUI.**Repo tools:** read and search code, create files, apply precise edits, and inspect session changes/diffs.**Terminal execution:** run shell commands with streamed output and project-level permission controls.**Plan/build workflow:** use read-only Plan mode for investigation and a reviewable task list, then Build mode to implement.**Web search and browsing:** DuckDuckGo works by default with no key; Firecrawl, Tavily, and SearXNG are configurable search backends. Kolega Code can also fetch URLs directly and use a Playwright-powered browser agent for interactive sites.**MCP servers:** connect verified`streamable_http`\n\n,`sse`\n\n, and`stdio`\n\nMCP servers (including OAuth-enabled HTTP servers) as permission-gated tools.**Model routing:** choose provider/model combinations, set thinking effort, split long-context/fast/thinking roles, and override models per agent role.**Interactive or scriptable:** use the Textual TUI, queue follow-up prompts while the agent is working, run`kolega-code ask`\n\n, request JSON output, list/export/resume sessions, and diagnose setup with`doctor`\n\n.**Extensibility:** add agent skills, override prompts with project templates, run lifecycle hooks, and persist project permission rules.**Local-first state:** sessions, settings, permissions, OAuth tokens, and API-key settings stay on your machine with restrictive permissions where applicable.\n\n**1. Install** with the script:\n\n```\ncurl -fsSL https://kolega.dev/install-kolega-code.sh | sh\n```\n\nOr with [uv](https://docs.astral.sh/uv/) (or `pip`\n\n):\n\n```\nuv tool install kolega-code\n# or: pip install kolega-code\n```\n\nVerify the install:\n\n```\nkolega-code --version\n```\n\n**2. Start a session** in your project:\n\n```\nkolega-code .\n```\n\n**3. Connect a model.** Complete the first-run wizard with ChatGPT sign-in or an\nAPI key. Later, open the categorized Settings screen from the sidebar or with\n`/settings`\n\n. Use `Shift+Tab`\n\nto switch between **Plan** and **Build** mode, or run\n`/gigacode on`\n\nwhen a task is broad enough for fan-out.\n\nResume a previous conversation:\n\n```\nkolega-code . --resume            # latest session\nkolega-code . --resume <id>       # a specific Resume ID from `sessions list`\n```\n\n| Mode | Command | Best for |\n|---|---|---|\nInteractive TUI |\n`kolega-code .` |\nDay-to-day development, exploration, orchestration |\nOne-shot |\n`kolega-code ask \"…\"` |\nScripting, automation, quick questions, CI |\n\nThere are also helper commands for managing sessions and checking your setup:\n\n```\nkolega-code ask \"summarize this repository\" --project .\nkolega-code sessions list --project .\nkolega-code doctor --project .\n```\n\nKolega Code supports a broad model-provider catalog and lets you route models by role instead of forcing one model to do every job.\n\nSupported model providers:\n\n- Anthropic\n- OpenAI API\n- OpenAI via ChatGPT subscription sign-in\n- Groq\n- Together.ai\n- Fireworks.ai\n- xAI / Grok\n- DashScope / Qwen\n- Moonshot / Kimi\n- DeepSeek\n- Z.AI / GLM Coding Plan\n- Kimi Coding Plan\n- Ollama Cloud\n- local Llama\n\nSupported web-search backends:\n\n- DuckDuckGo — default, no key required\n- Firecrawl\n- Tavily\n- SearXNG — self-hosted option\n\nSee [Providers & Models](https://kolega-ai.github.io/kolega-code/configuration/providers-and-models/)\nfor model IDs, role configuration, API-key variables, and thinking-effort options.\n\nKolega Code can assign different models to different operational roles: a strong long-context model for the main coding loop, a faster model for utility calls, and a dedicated model for extended thinking. You can also override models per agent role — planning, building, investigation, general, and browser — so wide workflows can use cheaper models where they fit and stronger models where they matter.\n\nIf you have a paid **ChatGPT** plan (Plus, Pro, or Business), you can use it to run\nOpenAI models instead of a separate API key. Run `/login chatgpt`\n\nin the TUI,\ncomplete the browser sign-in, and Kolega Code switches to the **OpenAI (ChatGPT\nsubscription)** provider (default `gpt-5.6-sol`\n\n). Tokens are stored locally (chmod\n`600`\n\n) and refreshed automatically; `/logout chatgpt`\n\nremoves them. See\n[Sign in with ChatGPT](https://kolega-ai.github.io/kolega-code/configuration/sign-in-with-chatgpt/).\n\nSet your provider, model, and API keys from first-run onboarding or the full-screen Settings editor in the UI, or via environment variables and flags for non-interactive use:\n\n```\nexport KOLEGA_CODE_PROVIDER=deepseek\nexport DEEPSEEK_API_KEY=...\nkolega-code ask \"summarize this repository\" --project . --provider deepseek --model deepseek-v4-pro\n```\n\nAPI key variables only provide credentials — pick a provider/model explicitly or\nsave one in Settings. Local session state lives under your platform's state\ndirectory unless `KOLEGA_CODE_STATE_DIR`\n\nis set. See the\n[Configuration docs](https://kolega-ai.github.io/kolega-code/configuration/settings-and-api-keys/)\nfor the full story.\n\nThe `web_search`\n\ntool uses DuckDuckGo by default without a key. To choose another\nbackend, set it in **Settings** or export `KOLEGA_CODE_WEB_SEARCH_BACKEND`\n\nas\n`firecrawl`\n\n, `tavily`\n\n, or `searxng`\n\n; use `FIRECRAWL_API_KEY`\n\n, `TAVILY_API_KEY`\n\n, or\n`SEARXNG_BASE_URL`\n\nas needed.\n\nProjects can override Kolega Code's base prompts with uppercase Markdown templates\nin `.kolega/prompts/`\n\n. Generate editable starters with Jinja replacement tags using\n`/prompts dump`\n\nin the TUI or `kolega-code prompts dump --project .`\n\nin a terminal.\nTo dump only selected starters, pass prompt names such as `coder`\n\n, `planning`\n\n, or\n`compaction`\n\n(filename aliases like `CODER.md`\n\nwork too). Validate existing\noverrides with `/prompts validate`\n\nor `kolega-code prompts validate --project .`\n\n.\n\n**Python 3.11+**- An\n**API key**, ChatGPT sign-in, or local model for at least one supported model provider - A terminal that supports a modern TUI (most do)\n\nFull documentation lives at ** kolega-ai.github.io/kolega-code**:\n\nContributions are welcome. See [CONTRIBUTING.md](/kolega-ai/kolega-code/blob/main/CONTRIBUTING.md) for local setup,\nrunning the test suite, and building the docs site. Please report security issues\nprivately per [SECURITY.md](/kolega-ai/kolega-code/blob/main/SECURITY.md).\n\nReleased under the [Apache License 2.0](/kolega-ai/kolega-code/blob/main/LICENSE).", "url": "https://wpnews.pro/news/show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli", "canonical_source": "https://github.com/kolega-ai/kolega-code", "published_at": "2026-07-24 10:33:37+00:00", "updated_at": "2026-07-24 10:52:24.170687+00:00", "lang": "en", "topics": ["developer-tools", "ai-agents", "ai-tools"], "entities": ["Kolega Code", "Kolega AI", "Gigacode", "DuckDuckGo", "Firecrawl", "Tavily", "SearXNG", "Playwright"], "alternates": {"html": "https://wpnews.pro/news/show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli", "markdown": "https://wpnews.pro/news/show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli.md", "text": "https://wpnews.pro/news/show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli.txt", "jsonld": "https://wpnews.pro/news/show-hn-frontier-model-pricing-became-a-rip-off-so-i-built-an-open-source-cli.jsonld"}}