research-assistant 0.1.0 → 0.1.1

package/README-ZH.md

@@ -86,7 +86,7 @@ base_url = "https://api.exa.ai"
 
 [[provider]]
 type     = "openai_compat"          # 驱动 `ask`
-base_url = "https://api.example.com/v1"
+base_url = "https://api.openai.com/v1"  # 默认；用兼容渠道时换成其地址（需含 /v1）
 api_key  = "..."
 model    = "grok-..."
 

package/README.md

@@ -93,7 +93,7 @@ base_url = "https://api.exa.ai"
 
 [[provider]]
 type     = "openai_compat"          # drives `ask`
-base_url = "https://api.example.com/v1"
+base_url = "https://api.openai.com/v1"  # default; for a compatible gateway, use its URL with /v1
 api_key  = "..."
 model    = "grok-..."
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "research-assistant",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "Search/research sub-agent + concurrent search/fetch/locate CLI atomic tools for AI agents (Exa, Tavily, Firecrawl, Context7, OpenAI-compatible).",
   "license": "MIT",
   "homepage": "https://github.com/arsonist-g/research-assistant-cli#readme",

package/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "research-assistant"
-version = "0.1.0"
+version = "0.1.1"
 description = "Search/research sub-agent + a set of concurrent search/fetch/locate CLI atomic tools for AI agents."
 readme = "README.md"
 requires-python = ">=3.10"

package/src/research_assistant/assets/skills/research-assistant/SKILL.md

@@ -97,7 +97,8 @@ Notes:
 
 `ctx7 library "<name>" "<question>"` resolves a library id. `ctx7 docs /org/repo
 "<question>"` returns up-to-date official docs as markdown. It hits the authoritative source
-directly, so it is fast and accurate.
+directly, so it is fast and accurate. On Windows git bash / WSL, MSYS rewrites the leading `/`
+of `/org/repo` into a Windows path; write `//org/repo` there (PowerShell and cmd are unaffected).
 
 **For any library, framework, SDK, CLI, or cloud-service question, prefer `ctx7` over web
 search.** Do not grep the open web for API syntax, config keys, or version-migration notes
@@ -168,6 +169,8 @@ research-assistant tavily map <url> [--max-depth N] [--max-breadth N] [--limit N
 research-assistant tavily crawl <url> [--max-depth N] [--limit N] [--instructions TEXT] [--extract-depth basic|advanced] [--no-external] [--timeout N]   # recursive crawl, markdown per page
 
 # Tier 2: Firecrawl. Residential bandwidth. scrape and search are keyless; the rest need a key.
+# Keyless endpoints are sensitive to IP quality: residential/home IPs work; datacenter IPs may
+# hit 403 ("IP looks suspicious") and need a home-bandwidth proxy.
 research-assistant firecrawl scrape <url>... [--format markdown|html] [--only-main-content] [--wait-for MS]   # batch markdown, keyless
 research-assistant firecrawl search "<query>" [--limit N] [--sources web|news] [--scrape]                      # keyless; --scrape returns full markdown per result
 research-assistant firecrawl map <url> [--limit N] [--include-subdomains]                                      # needs key
@@ -194,7 +197,7 @@ research-assistant search "<query>" [--providers exa,tavily,firecrawl,browser] [
 
 ```sh
 research-assistant ctx7 library "<name>" "<question>"       # resolve a library id
-research-assistant ctx7 docs /org/repo "<question>"         # up-to-date official docs
+research-assistant ctx7 docs /org/repo "<question>"         # up-to-date official docs (git bash/WSL: use //org/repo)
 ```
 
 ### C. Ask a web-enabled LLM in natural language (`ask`)

package/src/research_assistant/commands/doctor.py

@@ -143,6 +143,21 @@ def _minimal_namespace(cap_name: str) -> argparse.Namespace:
         "ids": [],
         "url": "https://example.com",
         "urls": ["https://example.com"],
+        # search provider 完整字段：exa/tavily 等 handler 直接访问 ns.<attr>，缺就 AttributeError。
+        # 这里给全 superset，doctor 测任何 provider 的 search 能力都不缺字段。
+        "end_date": None,
+        "include_text": None,
+        "exclude_text": None,
+        "include_raw_content": None,
+        "chunks_per_source": None,
+        "country": None,
+        "days": None,
+        "include_images": False,
+        "include_image_descriptions": False,
+        "include_favicon": False,
+        "auto_parameters": False,
+        "engine": "bing-intl",
+        "max_pages": 10,
     }
     return argparse.Namespace(**base)
 

package/src/research_assistant/providers/openai_compat.py

@@ -42,8 +42,9 @@ def resolve_cfg(config: Config, provider_type: str = "openai_compat") -> Provide
     cfg = config.require_provider(provider_type)
     if not cfg.api_key:
         raise ArgsError(f"{provider_type}: 缺少 api_key", provider=provider_type)
+    # base_url 默认 OpenAI 官方（含 /v1/）；用 OpenAI 兼容渠道时 base_url 需自带 /v1/（如 https://host/v1）
     if not cfg.base_url:
-        raise ArgsError(f"{provider_type}: 缺少 base_url", provider=provider_type)
+        cfg.base_url = "https://api.openai.com/v1"
     if not cfg.model:
         raise ArgsError(f"{provider_type}: 缺少 model", provider=provider_type)
     return cfg
@@ -107,8 +108,12 @@ async def _stream_chat(client: httpx.AsyncClient, body: dict[str, Any], provider
             search_results: list[Any] = []
             usage: dict[str, Any] | None = None
             role = "assistant"
+            raw_non_sse: list[str] = []  # 非 data: 行：网关可能 HTTP 200 却返回 body JSON error（非 SSE）
             async for line in resp.aiter_lines():
-                if not line or not line.startswith("data:"):
+                if not line:
+                    continue
+                if not line.startswith("data:"):
+                    raw_non_sse.append(line)
                     continue
                 data_str = line[5:].strip()
                 if data_str == "[DONE]":
@@ -141,6 +146,25 @@ async def _stream_chat(client: httpx.AsyncClient, body: dict[str, Any], provider
 
         raise wrap_provider_http_error(provider_type, e) from e
 
+    # 网关 HTTP 200 却没给 SSE chunk：body 可能是 JSON error 或 HTML 错误页（上游不可用等）。
+    # 别静默返回空 content（否则用户以为 research-assistant 坏，实际是 model 上游）。
+    if not content_parts and raw_non_sse:
+        raw = "\n".join(raw_non_sse)
+        msg: str | None = None
+        try:
+            err_body = json.loads(raw)
+            if isinstance(err_body, dict) and err_body.get("error"):
+                err = err_body["error"]
+                msg = err.get("message") if isinstance(err, dict) else str(err)
+        except json.JSONDecodeError:
+            pass
+        if msg:
+            raise ProviderError(f"{provider_type}: 上游返回错误 — {msg}", provider=provider_type)
+        raise ProviderError(
+            f"{provider_type}: 上游返回非 SSE 响应（HTTP 200，可能错误页/上游不可用）: {raw[:120]}",
+            provider=provider_type,
+        )
+
     message: dict[str, Any] = {"role": role, "content": "".join(content_parts)}
     if citations:
         message["citations"] = citations