1 line
12 KiB
JSON
1 line
12 KiB
JSON
{"repo": "bojun-vvibe/oss-contributions", "files": {"reviews/2026-W17/drip-85/Aider-AI-aider-pr-5069.md": {"73f30d133053141afef1f5158a6204841a529da6": {"t": 1785747230.1514199, "n": 5853, "h": "4850481862fd1f54", "c": "# Aider-AI/aider PR #5069 — docs: Add FuturMix AI Gateway to LLM providers\n\n- **Repo:** Aider-AI/aider\n- **PR:** [#5069](https://github.com/Aider-AI/aider/pull/5069)\n- **Files:** 1 file, +77/−0 (new file `aider/website/docs/llms/futurmix.md`)\n\n## Context\n\nAdds a new docs page under \"Connecting to LLMs\" describing FuturMix\n(an OpenAI-compatible AI gateway claiming 22+ models from\nAnthropic/OpenAI/Google/DeepSeek/xAI behind one endpoint at\n`https://futurmix.ai/v1`). The page follows the same template as\nthe existing per-provider docs (`openrouter.md`, `together.md`, etc.):\nfront-matter with `parent: Connecting to LLMs`, install snippet,\nenv-var export, sample `aider --model …` invocations, model list.\n\n## What the diff actually does\n\n**`aider/website/docs/llms/futurmix.md` (new, 77 lines)**:\n- Front-matter at `:1-4` with `nav_order: 500` (a high number, so it\n sorts toward the end of the LLM provider list).\n- Connection guidance at `:24-33`: export `OPENAI_API_BASE` and\n `OPENAI_API_KEY` (uses the OpenAI-compatible path, no new\n litellm provider prefix).\n- Sample invocations at `:42-48`:\n - `aider --model openai/claude-sonnet-4-20250514`\n - `aider --model openai/gpt-4-turbo`\n - `aider --model openai/gemini-2.0-flash-exp`\n- Curated model list at `:55-73` split into Anthropic / OpenAI /\n Google sections.\n- Closing \"Benefits\" bulletlist at `:79-83` repeating the marketing\n claims (single API key, OpenAI-compatible, 99.99% SLA,\n competitive pricing, easy switching).\n\n## Strengths\n\n- **Follows the template**. Front-matter shape, `{% include\n install.md %}` shortcode, and Mac/Linux/Windows env-var pattern\n all match the existing per-provider pages, so the page slots in\n cleanly without theme work.\n- **OpenAI-compatible path is the right approach** — no new\n litellm provider prefix is required because FuturMix exposes\n the `/v1` shape, so `OPENAI_API_BASE` redirection works out of\n the box. No code changes needed in aider itself.\n- **Concrete invocations at `:42-48`** give the user a paste-and-go\n starting point rather than just abstract \"set these env vars.\"\n\n## Risks / nits\n\n1. **Marketing claims unverified.** \"99.99% SLA\", \"22+ AI models\",\n \"competitive pricing\" all appear at `:14` and `:79-82` without\n citation. Aider's docs aren't the place to repeat a vendor's\n marketing copy verbatim — the maintainers will have to choose\n between (a) trimming to neutral facts (\"OpenAI-compatible\n gateway exposing models from multiple providers\") or (b)\n accepting the precedent that vendor-submitted docs PRs may\n carry promotional language. Recommend (a).\n2. **`nav_order: 500` is suspicious.** A quick check of other\n provider pages will probably show `nav_order` values in the\n 1-50 range. 500 will work (Jekyll just sorts numerically) but\n it suggests the author didn't read the conventions. Pick a\n number consistent with neighbors (probably alphabetical\n placement between `fireworks.md` and `gemini.md`).\n3. **Model id `openai/claude-sonnet-4-20250514`** at `:42` is\n semantically odd — it's a Claude model with the `openai/`\n litellm prefix because the gateway is OpenAI-compatible. This\n *will* work but reads as a typo. Add a one-sentence callout:\n \"Note: all models use the `openai/` prefix because FuturMix\n speaks the OpenAI API, regardless of the underlying provider.\"\n4. **`gemini-2.0-flash-exp` is an experimental ID.** By the time\n this doc lands, the `-exp` suffix may have been replaced with\n a stable id. Either link to FuturMix's model list (already\n done at `:75`) and prune the inline list, or add a \"as of\n date\" timestamp.\n5. **No verification that FuturMix actually works with aider.**\n Docs PRs for niche gateways have historically landed without\n anyone on the maintainer side actually testing the path. Ask\n the author to attach a screenshot or transcript of `aider\n --model openai/<some-model>` connecting successfully through\n FuturMix.\n6. **The \"Benefits\" section at `:77-83`** is pure marketing and\n adds no operational value. Recommend deletion — the user\n already chose to read this page, the sales pitch is redundant.\n7. **No mention of rate limits, error handling, or what happens\n when a model is unavailable on the gateway.** Existing\n provider pages (e.g., `openrouter.md`) typically include a\n troubleshooting hint. A \"if you see X, try Y\" line would\n match precedent.\n8. **Author appears to be FuturMix-affiliated** (not verified in\n the metadata, but vendor-authored docs PRs are common). That's\n not disqualifying — most provider docs in `aider/website/docs/llms/`\n were originally vendor-submitted — but it does heighten the\n bar for trimming marketing language.\n9. **No CHANGELOG / release-notes entry.** Adding a provider page\n typically gets a one-line mention in `HISTORY.md`. Worth\n confirming with maintainers whether this is required for docs\n PRs.\n\n## Verdict\n\n`request-changes`\n\nThe page is structurally correct (template followed, front-matter\nvalid, OpenAI-compatible path is the right integration approach)\nbut it reads as vendor-supplied marketing copy rather than\noperational docs. Three concrete asks before merge:\n\n1. Strip the \"99.99% SLA / competitive pricing / Cost Effective\"\n marketing language at `:14` and `:77-83` down to neutral\n factual statements.\n2. Fix `nav_order: 500` to something consistent with neighboring\n provider pages.\n3. Add a one-sentence note explaining why Claude/Gemini models\n use the `openai/` litellm prefix (because the gateway is\n OpenAI-compatible), so users aren't confused by\n `openai/claude-sonnet-4-20250514`.\n\nOptional: attach proof-of-life (`aider` actually connecting through\nFuturMix) and add a \"tested as of YYYY-MM-DD\" footer so the model\nlist has a known staleness window. Once those land it's a clean\n`merge-as-is`.\n"}, "3d6eafc997b6b3bfb49eddd10ca42800ab675d27": {"t": 1785771536.3429892, "n": 5853, "h": "4850481862fd1f54", "c": "# Aider-AI/aider PR #5069 — docs: Add FuturMix AI Gateway to LLM providers\n\n- **Repo:** Aider-AI/aider\n- **PR:** [#5069](https://github.com/Aider-AI/aider/pull/5069)\n- **Files:** 1 file, +77/−0 (new file `aider/website/docs/llms/futurmix.md`)\n\n## Context\n\nAdds a new docs page under \"Connecting to LLMs\" describing FuturMix\n(an OpenAI-compatible AI gateway claiming 22+ models from\nAnthropic/OpenAI/Google/DeepSeek/xAI behind one endpoint at\n`https://futurmix.ai/v1`). The page follows the same template as\nthe existing per-provider docs (`openrouter.md`, `together.md`, etc.):\nfront-matter with `parent: Connecting to LLMs`, install snippet,\nenv-var export, sample `aider --model …` invocations, model list.\n\n## What the diff actually does\n\n**`aider/website/docs/llms/futurmix.md` (new, 77 lines)**:\n- Front-matter at `:1-4` with `nav_order: 500` (a high number, so it\n sorts toward the end of the LLM provider list).\n- Connection guidance at `:24-33`: export `OPENAI_API_BASE` and\n `OPENAI_API_KEY` (uses the OpenAI-compatible path, no new\n litellm provider prefix).\n- Sample invocations at `:42-48`:\n - `aider --model openai/claude-sonnet-4-20250514`\n - `aider --model openai/gpt-4-turbo`\n - `aider --model openai/gemini-2.0-flash-exp`\n- Curated model list at `:55-73` split into Anthropic / OpenAI /\n Google sections.\n- Closing \"Benefits\" bulletlist at `:79-83` repeating the marketing\n claims (single API key, OpenAI-compatible, 99.99% SLA,\n competitive pricing, easy switching).\n\n## Strengths\n\n- **Follows the template**. Front-matter shape, `{% include\n install.md %}` shortcode, and Mac/Linux/Windows env-var pattern\n all match the existing per-provider pages, so the page slots in\n cleanly without theme work.\n- **OpenAI-compatible path is the right approach** — no new\n litellm provider prefix is required because FuturMix exposes\n the `/v1` shape, so `OPENAI_API_BASE` redirection works out of\n the box. No code changes needed in aider itself.\n- **Concrete invocations at `:42-48`** give the user a paste-and-go\n starting point rather than just abstract \"set these env vars.\"\n\n## Risks / nits\n\n1. **Marketing claims unverified.** \"99.99% SLA\", \"22+ AI models\",\n \"competitive pricing\" all appear at `:14` and `:79-82` without\n citation. Aider's docs aren't the place to repeat a vendor's\n marketing copy verbatim — the maintainers will have to choose\n between (a) trimming to neutral facts (\"OpenAI-compatible\n gateway exposing models from multiple providers\") or (b)\n accepting the precedent that vendor-submitted docs PRs may\n carry promotional language. Recommend (a).\n2. **`nav_order: 500` is suspicious.** A quick check of other\n provider pages will probably show `nav_order` values in the\n 1-50 range. 500 will work (Jekyll just sorts numerically) but\n it suggests the author didn't read the conventions. Pick a\n number consistent with neighbors (probably alphabetical\n placement between `fireworks.md` and `gemini.md`).\n3. **Model id `openai/claude-sonnet-4-20250514`** at `:42` is\n semantically odd — it's a Claude model with the `openai/`\n litellm prefix because the gateway is OpenAI-compatible. This\n *will* work but reads as a typo. Add a one-sentence callout:\n \"Note: all models use the `openai/` prefix because FuturMix\n speaks the OpenAI API, regardless of the underlying provider.\"\n4. **`gemini-2.0-flash-exp` is an experimental ID.** By the time\n this doc lands, the `-exp` suffix may have been replaced with\n a stable id. Either link to FuturMix's model list (already\n done at `:75`) and prune the inline list, or add a \"as of\n date\" timestamp.\n5. **No verification that FuturMix actually works with aider.**\n Docs PRs for niche gateways have historically landed without\n anyone on the maintainer side actually testing the path. Ask\n the author to attach a screenshot or transcript of `aider\n --model openai/<some-model>` connecting successfully through\n FuturMix.\n6. **The \"Benefits\" section at `:77-83`** is pure marketing and\n adds no operational value. Recommend deletion — the user\n already chose to read this page, the sales pitch is redundant.\n7. **No mention of rate limits, error handling, or what happens\n when a model is unavailable on the gateway.** Existing\n provider pages (e.g., `openrouter.md`) typically include a\n troubleshooting hint. A \"if you see X, try Y\" line would\n match precedent.\n8. **Author appears to be FuturMix-affiliated** (not verified in\n the metadata, but vendor-authored docs PRs are common). That's\n not disqualifying — most provider docs in `aider/website/docs/llms/`\n were originally vendor-submitted — but it does heighten the\n bar for trimming marketing language.\n9. **No CHANGELOG / release-notes entry.** Adding a provider page\n typically gets a one-line mention in `HISTORY.md`. Worth\n confirming with maintainers whether this is required for docs\n PRs.\n\n## Verdict\n\n`request-changes`\n\nThe page is structurally correct (template followed, front-matter\nvalid, OpenAI-compatible path is the right integration approach)\nbut it reads as vendor-supplied marketing copy rather than\noperational docs. Three concrete asks before merge:\n\n1. Strip the \"99.99% SLA / competitive pricing / Cost Effective\"\n marketing language at `:14` and `:77-83` down to neutral\n factual statements.\n2. Fix `nav_order: 500` to something consistent with neighboring\n provider pages.\n3. Add a one-sentence note explaining why Claude/Gemini models\n use the `openai/` litellm prefix (because the gateway is\n OpenAI-compatible), so users aren't confused by\n `openai/claude-sonnet-4-20250514`.\n\nOptional: attach proof-of-life (`aider` actually connecting through\nFuturMix) and add a \"tested as of YYYY-MM-DD\" footer so the model\nlist has a known staleness window. Once those land it's a clean\n`merge-as-is`.\n"}}}} |