kavel 0.1.0__tar.gz

kavel-0.1.0/PKG-INFO

@@ -0,0 +1,83 @@
+Metadata-Version: 2.4
+Name: kavel
+Version: 0.1.0
+Summary: Generate images from a prompt with no API key and no account
+Author: hanshs474
+License: MIT
+Project-URL: Homepage, https://www.kavel.ai/?ref=pypi
+Project-URL: Image tools, https://www.kavel.ai/image?ref=pypi
+Project-URL: Video tools, https://www.kavel.ai/video?ref=pypi
+Project-URL: Source, https://github.com/hanshs474/kavel-image-skill
+Keywords: ai,image-generation,text-to-image,image-editing,no-api-key
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Multimedia :: Graphics
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+
+# kavel
+
+Generate images from a prompt with **no API key and no account**.
+
+```python
+from kavel import generate
+
+url = generate("a paper-cut layered mountain range at dusk, warm rim light")
+```
+
+That is the whole setup. No key to fetch, no config file, no dependencies —
+standard library only. It calls the anonymous endpoint on
+[Kavel](https://www.kavel.ai/?ref=pypi), an online AI image and video studio.
+
+## Editing a photo
+
+Pass `image_url` and the prompt describes the change rather than the picture:
+
+```python
+url = generate(
+    "restyle the hair into a shoulder-length layered cut, "
+    "keep the same face, skin and lighting",
+    image_url="https://example.com/portrait.jpg",
+)
+```
+
+Naming what must stay is what holds the likeness. Prompts that only name the
+edit tend to drift the whole face.
+
+## What the free tier actually covers
+
+| | |
+|---|---|
+| Anonymous grant | 15 credits |
+| Text-to-image | 5 credits — so the grant covers two |
+| Photo edit (`nano-banana-2-lite`) | 15 credits — exactly one |
+| Per-IP ceiling | 30 credits a day |
+
+Anonymous output is 1K and watermarked. Signing in on
+[kavel.ai](https://www.kavel.ai/?ref=pypi) removes the watermark and opens the
+larger models.
+
+**Video is deliberately not exposed here.** The cheapest clip is 40 credits
+against a 15-credit anonymous wallet, so a signed-out call cannot succeed at
+any setting. Signed-out video on the site runs on a browser-side engine
+instead — see [the video tools](https://www.kavel.ai/video?ref=pypi).
+
+## Errors worth catching
+
+`KavelError` covers the three things that actually happen:
+
+- the content filter refused the prompt (credits are spent either way, so
+  rewrite rather than retry)
+- the queue ran past `timeout`
+- the per-IP daily ceiling returned 429
+
+## Prompts that work
+
+Kavel keeps a page per edit, each preloaded with a prompt for that specific
+job — [hairstyle](https://www.kavel.ai/image/ai-hairstyle-changer?ref=pypi),
+[outfit](https://www.kavel.ai/image/ai-outfit-generator?ref=pypi),
+and the rest under [image tools](https://www.kavel.ai/image?ref=pypi).
+
+MIT licensed.

kavel-0.1.0/README.md

@@ -0,0 +1,64 @@
+# kavel
+
+Generate images from a prompt with **no API key and no account**.
+
+```python
+from kavel import generate
+
+url = generate("a paper-cut layered mountain range at dusk, warm rim light")
+```
+
+That is the whole setup. No key to fetch, no config file, no dependencies —
+standard library only. It calls the anonymous endpoint on
+[Kavel](https://www.kavel.ai/?ref=pypi), an online AI image and video studio.
+
+## Editing a photo
+
+Pass `image_url` and the prompt describes the change rather than the picture:
+
+```python
+url = generate(
+    "restyle the hair into a shoulder-length layered cut, "
+    "keep the same face, skin and lighting",
+    image_url="https://example.com/portrait.jpg",
+)
+```
+
+Naming what must stay is what holds the likeness. Prompts that only name the
+edit tend to drift the whole face.
+
+## What the free tier actually covers
+
+| | |
+|---|---|
+| Anonymous grant | 15 credits |
+| Text-to-image | 5 credits — so the grant covers two |
+| Photo edit (`nano-banana-2-lite`) | 15 credits — exactly one |
+| Per-IP ceiling | 30 credits a day |
+
+Anonymous output is 1K and watermarked. Signing in on
+[kavel.ai](https://www.kavel.ai/?ref=pypi) removes the watermark and opens the
+larger models.
+
+**Video is deliberately not exposed here.** The cheapest clip is 40 credits
+against a 15-credit anonymous wallet, so a signed-out call cannot succeed at
+any setting. Signed-out video on the site runs on a browser-side engine
+instead — see [the video tools](https://www.kavel.ai/video?ref=pypi).
+
+## Errors worth catching
+
+`KavelError` covers the three things that actually happen:
+
+- the content filter refused the prompt (credits are spent either way, so
+  rewrite rather than retry)
+- the queue ran past `timeout`
+- the per-IP daily ceiling returned 429
+
+## Prompts that work
+
+Kavel keeps a page per edit, each preloaded with a prompt for that specific
+job — [hairstyle](https://www.kavel.ai/image/ai-hairstyle-changer?ref=pypi),
+[outfit](https://www.kavel.ai/image/ai-outfit-generator?ref=pypi),
+and the rest under [image tools](https://www.kavel.ai/image?ref=pypi).
+
+MIT licensed.

kavel-0.1.0/pyproject.toml

@@ -0,0 +1,29 @@
+[build-system]
+requires = ["setuptools>=68"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "kavel"
+version = "0.1.0"
+description = "Generate images from a prompt with no API key and no account"
+readme = "README.md"
+requires-python = ">=3.8"
+license = { text = "MIT" }
+authors = [{ name = "hanshs474" }]
+keywords = ["ai", "image-generation", "text-to-image", "image-editing", "no-api-key"]
+classifiers = [
+  "Development Status :: 4 - Beta",
+  "Intended Audience :: Developers",
+  "License :: OSI Approved :: MIT License",
+  "Programming Language :: Python :: 3",
+  "Topic :: Multimedia :: Graphics",
+]
+
+[project.urls]
+Homepage = "https://www.kavel.ai/?ref=pypi"
+"Image tools" = "https://www.kavel.ai/image?ref=pypi"
+"Video tools" = "https://www.kavel.ai/video?ref=pypi"
+Source = "https://github.com/hanshs474/kavel-image-skill"
+
+[tool.setuptools.packages.find]
+where = ["src"]

kavel-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

kavel-0.1.0/src/kavel/__init__.py

@@ -0,0 +1,132 @@
+"""Generate images from a prompt with no API key and no account.
+
+Calls the anonymous endpoint on kavel.ai — https://www.kavel.ai/?ref=pypi
+
+    from kavel import generate
+    url = generate("a paper-cut layered mountain range at dusk")
+
+Standard library only. No dependencies, no configuration.
+"""
+
+from __future__ import annotations
+
+import json
+import time
+import urllib.error
+import urllib.request
+import uuid
+
+__version__ = "0.1.0"
+__all__ = ["generate", "KavelError", "BASE"]
+
+BASE = "https://www.kavel.ai"
+
+#: Anonymous wallet, measured against the live site on 2026-09-04.
+#: A text-to-image run costs 5 of the 15 credits a fresh visitor is granted;
+#: an edit (``nano-banana-2-lite``) costs all 15. There is also a per-IP
+#: ceiling of 30 credits a day. Video cannot be paid for anonymously at any
+#: setting, which is why this module does not expose it.
+ANON_GRANT = 15
+COST_TEXT_TO_IMAGE = 5
+COST_EDIT = 15
+
+
+class KavelError(RuntimeError):
+    """Raised when a generation is refused, times out, or comes back empty."""
+
+
+def _post(path: str, payload: dict, anon: str) -> dict:
+    req = urllib.request.Request(
+        BASE + path,
+        data=json.dumps(payload).encode(),
+        headers={
+            "Content-Type": "application/json",
+            "x-anon-id": anon,
+            "User-Agent": f"kavel-python/{__version__}",
+        },
+    )
+    try:
+        with urllib.request.urlopen(req, timeout=60) as r:
+            return json.loads(r.read())
+    except urllib.error.HTTPError as e:
+        if e.code == 429:
+            raise KavelError(
+                "per-IP daily ceiling reached (30 credits); wait or sign in"
+            ) from e
+        raise KavelError(f"HTTP {e.code}: {e.read()[:200]!r}") from e
+
+
+def _get(path: str, anon: str) -> dict:
+    req = urllib.request.Request(
+        BASE + path,
+        headers={"x-anon-id": anon, "User-Agent": f"kavel-python/{__version__}"},
+    )
+    with urllib.request.urlopen(req, timeout=60) as r:
+        return json.loads(r.read())
+
+
+def generate(
+    prompt: str,
+    *,
+    aspect_ratio: str = "1:1",
+    image_url: str | None = None,
+    timeout: float = 240.0,
+    poll_every: float = 4.0,
+) -> str:
+    """Generate one image and return its CDN URL.
+
+    ``prompt`` describes the picture. Pass ``image_url`` to edit an existing
+    photo instead of generating from scratch — the prompt then describes the
+    change and should name what must stay ("keep the same face and lighting"),
+    because anything left unsaid is fair game for the model.
+
+    Editing runs on ``nano-banana-2-lite`` and costs the whole anonymous grant,
+    so a signed-out caller gets one edit. Text-to-image runs on
+    ``kavel-image-v1``, which takes no image input.
+
+    Raises :class:`KavelError` on refusal, timeout, or an empty result.
+    """
+    anon = f"anon-{uuid.uuid4()}"
+    model = "nano-banana-2-lite" if image_url else "kavel-image-v1"
+    scene = "image-to-image" if image_url else "text-to-image"
+    options: dict = {"aspect_ratio": aspect_ratio}
+    if image_url:
+        options["imageUrl"] = image_url
+
+    submitted = _post(
+        "/api/ai/generate",
+        {
+            "provider": "kie",
+            "mediaType": "image",
+            "model": model,
+            "scene": scene,
+            "prompt": prompt,
+            "options": options,
+        },
+        anon,
+    )
+    task_id = (submitted.get("data") or {}).get("id")
+    if not task_id:
+        raise KavelError(f"no task id in response: {str(submitted)[:200]}")
+
+    deadline = time.monotonic() + timeout
+    while time.monotonic() < deadline:
+        time.sleep(poll_every)
+        data = _get(
+            f"/api/ai/anon-query?taskId={task_id}"
+            "&provider=kie&mediaType=image",
+            anon,
+        ).get("data") or {}
+        images = data.get("images") or []
+        if images:
+            return images[0]
+        status = str(data.get("status") or "").lower()
+        if status in {"failed", "error"}:
+            # A finished task with no image is the content filter, and the
+            # credits are already spent — retrying the same text spends them
+            # again for the same answer.
+            raise KavelError(
+                "generation refused (content filter); rewrite the prompt "
+                "rather than retrying it"
+            )
+    raise KavelError(f"timed out after {timeout:g}s; the queue can run long")

kavel-0.1.0/src/kavel.egg-info/PKG-INFO

@@ -0,0 +1,83 @@
+Metadata-Version: 2.4
+Name: kavel
+Version: 0.1.0
+Summary: Generate images from a prompt with no API key and no account
+Author: hanshs474
+License: MIT
+Project-URL: Homepage, https://www.kavel.ai/?ref=pypi
+Project-URL: Image tools, https://www.kavel.ai/image?ref=pypi
+Project-URL: Video tools, https://www.kavel.ai/video?ref=pypi
+Project-URL: Source, https://github.com/hanshs474/kavel-image-skill
+Keywords: ai,image-generation,text-to-image,image-editing,no-api-key
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Multimedia :: Graphics
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+
+# kavel
+
+Generate images from a prompt with **no API key and no account**.
+
+```python
+from kavel import generate
+
+url = generate("a paper-cut layered mountain range at dusk, warm rim light")
+```
+
+That is the whole setup. No key to fetch, no config file, no dependencies —
+standard library only. It calls the anonymous endpoint on
+[Kavel](https://www.kavel.ai/?ref=pypi), an online AI image and video studio.
+
+## Editing a photo
+
+Pass `image_url` and the prompt describes the change rather than the picture:
+
+```python
+url = generate(
+    "restyle the hair into a shoulder-length layered cut, "
+    "keep the same face, skin and lighting",
+    image_url="https://example.com/portrait.jpg",
+)
+```
+
+Naming what must stay is what holds the likeness. Prompts that only name the
+edit tend to drift the whole face.
+
+## What the free tier actually covers
+
+| | |
+|---|---|
+| Anonymous grant | 15 credits |
+| Text-to-image | 5 credits — so the grant covers two |
+| Photo edit (`nano-banana-2-lite`) | 15 credits — exactly one |
+| Per-IP ceiling | 30 credits a day |
+
+Anonymous output is 1K and watermarked. Signing in on
+[kavel.ai](https://www.kavel.ai/?ref=pypi) removes the watermark and opens the
+larger models.
+
+**Video is deliberately not exposed here.** The cheapest clip is 40 credits
+against a 15-credit anonymous wallet, so a signed-out call cannot succeed at
+any setting. Signed-out video on the site runs on a browser-side engine
+instead — see [the video tools](https://www.kavel.ai/video?ref=pypi).
+
+## Errors worth catching
+
+`KavelError` covers the three things that actually happen:
+
+- the content filter refused the prompt (credits are spent either way, so
+  rewrite rather than retry)
+- the queue ran past `timeout`
+- the per-IP daily ceiling returned 429
+
+## Prompts that work
+
+Kavel keeps a page per edit, each preloaded with a prompt for that specific
+job — [hairstyle](https://www.kavel.ai/image/ai-hairstyle-changer?ref=pypi),
+[outfit](https://www.kavel.ai/image/ai-outfit-generator?ref=pypi),
+and the rest under [image tools](https://www.kavel.ai/image?ref=pypi).
+
+MIT licensed.

kavel-0.1.0/src/kavel.egg-info/SOURCES.txt

@@ -0,0 +1,7 @@
+README.md
+pyproject.toml
+src/kavel/__init__.py
+src/kavel.egg-info/PKG-INFO
+src/kavel.egg-info/SOURCES.txt
+src/kavel.egg-info/dependency_links.txt
+src/kavel.egg-info/top_level.txt

kavel-0.1.0/src/kavel.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

kavel-0.1.0/src/kavel.egg-info/top_level.txt

@@ -0,0 +1 @@
+kavel