hermes-agent/agent/image_gen_provider.py
Teknium ff9752410a
feat(plugins): pluggable image_gen backends + OpenAI provider (#13799)
* feat(plugins): pluggable image_gen backends + OpenAI provider

Adds a ImageGenProvider ABC so image generation backends register as
bundled plugins under `plugins/image_gen/<name>/`. The plugin scanner
gains three primitives to make this work generically:

- `kind:` manifest field (`standalone` | `backend` | `exclusive`).
  Bundled `kind: backend` plugins auto-load — no `plugins.enabled`
  incantation. User-installed backends stay opt-in.
- Path-derived keys: `plugins/image_gen/openai/` gets key
  `image_gen/openai`, so a future `tts/openai` cannot collide.
- Depth-2 recursion into category namespaces (parent dirs without a
  `plugin.yaml` of their own).

Includes `OpenAIImageGenProvider` as the first consumer (gpt-image-1.5
default, plus gpt-image-1, gpt-image-1-mini, DALL-E 3/2). Base64
responses save to `$HERMES_HOME/cache/images/`; URL responses pass
through.

FAL stays in-tree for this PR — a follow-up ports it into
`plugins/image_gen/fal/` so the in-tree `image_generation_tool.py`
slims down. The dispatch shim in `_handle_image_generate` only fires
when `image_gen.provider` is explicitly set to a non-FAL value, so
existing FAL setups are untouched.

- 41 unit tests (scanner recursion, kind parsing, gate logic,
  registry, OpenAI payload shapes)
- E2E smoke verified: bundled plugin autoloads, registers, and
  `_handle_image_generate` routes to OpenAI when configured

* fix(image_gen/openai): don't send response_format to gpt-image-*

The live API rejects it: 'Unknown parameter: response_format'
(verified 2026-04-21 with gpt-image-1.5). gpt-image-* models return
b64_json unconditionally, so the parameter was both unnecessary and
actively broken.

* feat(image_gen/openai): gpt-image-2 only, drop legacy catalog

gpt-image-2 is the latest/best OpenAI image model (released 2026-04-21)
and there's no reason to expose the older gpt-image-1.5 / gpt-image-1 /
dall-e-3 / dall-e-2 alongside it — slower, lower quality, or awkward
(dall-e-2 squares only). Trim the catalog down to a single model.

Live-verified end-to-end: landscape 1536x1024 render of a Moog-style
synth matches prompt exactly, 2.4MB PNG saved to cache.

* feat(image_gen/openai): expose gpt-image-2 as three quality tiers

Users pick speed/fidelity via the normal model picker instead of a
hidden quality knob. All three tier IDs resolve to the single underlying
gpt-image-2 API model with a different quality parameter:

  gpt-image-2-low     ~15s   fast iteration
  gpt-image-2-medium  ~40s   default
  gpt-image-2-high    ~2min  highest fidelity

Live-measured on OpenAI's API today: 15.4s / 40.8s / 116.9s for the
same 1024x1024 prompt.

Config:
  image_gen.openai.model: gpt-image-2-high
  # or
  image_gen.model: gpt-image-2-low
  # or env var for scripts/tests
  OPENAI_IMAGE_MODEL=gpt-image-2-medium

Live-verified end-to-end with the low tier: 18.8s landscape render of a
golden retriever in wildflowers, vision-confirmed exact match.

* feat(tools_config): plugin image_gen providers inject themselves into picker

'hermes tools' → Image Generation now shows plugin-registered backends
alongside Nous Subscription and FAL.ai without tools_config.py needing
to know about them. OpenAI appears as a third option today; future
backends appear automatically as they're added.

Mechanism:
- ImageGenProvider gains an optional get_setup_schema() hook
  (name, badge, tag, env_vars). Default derived from display_name.
- tools_config._plugin_image_gen_providers() pulls the schemas from
  every registered non-FAL plugin provider.
- _visible_providers() appends those rows when rendering the Image
  Generation category.
- _configure_provider() handles the new image_gen_plugin_name marker:
  writes image_gen.provider and routes to the plugin's list_models()
  catalog for the model picker.
- _toolset_needs_configuration_prompt('image_gen') stops demanding a
  FAL key when any plugin provider reports is_available().

FAL is skipped in the plugin path because it already has hardcoded
TOOL_CATEGORIES rows — when it gets ported to a plugin in a follow-up
PR the hardcoded rows go away and it surfaces through the same path
as OpenAI.

Verified live: picker shows Nous Subscription / FAL.ai / OpenAI.
Picking OpenAI prompts for OPENAI_API_KEY, then shows the
gpt-image-2-low/medium/high model picker sourced from the plugin.

397 tests pass across plugins/, tools_config, registry, and picker.

* fix(image_gen): close final gaps for plugin-backend parity with FAL

Two small places that still hardcoded FAL:

- hermes_cli/setup.py status line: an OpenAI-only setup showed
  'Image Generation: missing FAL_KEY'. Now probes plugin providers
  and reports '(OpenAI)' when one is_available() — or falls back to
  'missing FAL_KEY or OPENAI_API_KEY' if nothing is configured.

- image_generate tool schema description: said 'using FAL.ai, default
  FLUX 2 Klein 9B'. Rewrote provider-neutral — 'backend and model are
  user-configured' — and notes the 'image' field can be a URL or an
  absolute path, which the gateway delivers either way via
  extract_local_files().
2026-04-21 21:30:10 -07:00

242 lines
7.4 KiB
Python

"""
Image Generation Provider ABC
=============================
Defines the pluggable-backend interface for image generation. Providers register
instances via ``PluginContext.register_image_gen_provider()``; the active one
(selected via ``image_gen.provider`` in ``config.yaml``) services every
``image_generate`` tool call.
Providers live in ``<repo>/plugins/image_gen/<name>/`` (built-in, auto-loaded
as ``kind: backend``) or ``~/.hermes/plugins/image_gen/<name>/`` (user, opt-in
via ``plugins.enabled``).
Response shape
--------------
All providers return a dict that :func:`success_response` / :func:`error_response`
produce. The tool wrapper JSON-serializes it. Keys:
success bool
image str | None URL or absolute file path
model str provider-specific model identifier
prompt str echoed prompt
aspect_ratio str "landscape" | "square" | "portrait"
provider str provider name (for diagnostics)
error str only when success=False
error_type str only when success=False
"""
from __future__ import annotations
import abc
import base64
import datetime
import logging
import uuid
from pathlib import Path
from typing import Any, Dict, List, Optional, Tuple
logger = logging.getLogger(__name__)
VALID_ASPECT_RATIOS: Tuple[str, ...] = ("landscape", "square", "portrait")
DEFAULT_ASPECT_RATIO = "landscape"
# ---------------------------------------------------------------------------
# ABC
# ---------------------------------------------------------------------------
class ImageGenProvider(abc.ABC):
"""Abstract base class for an image generation backend.
Subclasses must implement :meth:`generate`. Everything else has sane
defaults — override only what your provider needs.
"""
@property
@abc.abstractmethod
def name(self) -> str:
"""Stable short identifier used in ``image_gen.provider`` config.
Lowercase, no spaces. Examples: ``fal``, ``openai``, ``replicate``.
"""
@property
def display_name(self) -> str:
"""Human-readable label shown in ``hermes tools``. Defaults to ``name.title()``."""
return self.name.title()
def is_available(self) -> bool:
"""Return True when this provider can service calls.
Typically checks for a required API key. Default: True
(providers with no external dependencies are always available).
"""
return True
def list_models(self) -> List[Dict[str, Any]]:
"""Return catalog entries for ``hermes tools`` model picker.
Each entry::
{
"id": "gpt-image-1.5", # required
"display": "GPT Image 1.5", # optional; defaults to id
"speed": "~10s", # optional
"strengths": "...", # optional
"price": "$...", # optional
}
Default: empty list (provider has no user-selectable models).
"""
return []
def get_setup_schema(self) -> Dict[str, Any]:
"""Return provider metadata for the ``hermes tools`` picker.
Used by ``tools_config.py`` to inject this provider as a row in
the Image Generation provider list. Shape::
{
"name": "OpenAI", # picker label
"badge": "paid", # optional short tag
"tag": "One-line description...", # optional subtitle
"env_vars": [ # keys to prompt for
{"key": "OPENAI_API_KEY",
"prompt": "OpenAI API key",
"url": "https://platform.openai.com/api-keys"},
],
}
Default: minimal entry derived from ``display_name``. Override to
expose API key prompts and custom badges.
"""
return {
"name": self.display_name,
"badge": "",
"tag": "",
"env_vars": [],
}
def default_model(self) -> Optional[str]:
"""Return the default model id, or None if not applicable."""
models = self.list_models()
if models:
return models[0].get("id")
return None
@abc.abstractmethod
def generate(
self,
prompt: str,
aspect_ratio: str = DEFAULT_ASPECT_RATIO,
**kwargs: Any,
) -> Dict[str, Any]:
"""Generate an image.
Implementations should return the dict from :func:`success_response`
or :func:`error_response`. ``kwargs`` may contain forward-compat
parameters future versions of the schema will expose — implementations
should ignore unknown keys.
"""
# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
def resolve_aspect_ratio(value: Optional[str]) -> str:
"""Clamp an aspect_ratio value to the valid set, defaulting to landscape.
Invalid values are coerced rather than rejected so the tool surface is
forgiving of agent mistakes.
"""
if not isinstance(value, str):
return DEFAULT_ASPECT_RATIO
v = value.strip().lower()
if v in VALID_ASPECT_RATIOS:
return v
return DEFAULT_ASPECT_RATIO
def _images_cache_dir() -> Path:
"""Return ``$HERMES_HOME/cache/images/``, creating parents as needed."""
from hermes_constants import get_hermes_home
path = get_hermes_home() / "cache" / "images"
path.mkdir(parents=True, exist_ok=True)
return path
def save_b64_image(
b64_data: str,
*,
prefix: str = "image",
extension: str = "png",
) -> Path:
"""Decode base64 image data and write it under ``$HERMES_HOME/cache/images/``.
Returns the absolute :class:`Path` to the saved file.
Filename format: ``<prefix>_<YYYYMMDD_HHMMSS>_<short-uuid>.<ext>``.
"""
raw = base64.b64decode(b64_data)
ts = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
short = uuid.uuid4().hex[:8]
path = _images_cache_dir() / f"{prefix}_{ts}_{short}.{extension}"
path.write_bytes(raw)
return path
def success_response(
*,
image: str,
model: str,
prompt: str,
aspect_ratio: str,
provider: str,
extra: Optional[Dict[str, Any]] = None,
) -> Dict[str, Any]:
"""Build a uniform success response dict.
``image`` may be an HTTP URL or an absolute filesystem path (for b64
providers like OpenAI). Callers that need to pass through additional
backend-specific fields can supply ``extra``.
"""
payload: Dict[str, Any] = {
"success": True,
"image": image,
"model": model,
"prompt": prompt,
"aspect_ratio": aspect_ratio,
"provider": provider,
}
if extra:
for k, v in extra.items():
payload.setdefault(k, v)
return payload
def error_response(
*,
error: str,
error_type: str = "provider_error",
provider: str = "",
model: str = "",
prompt: str = "",
aspect_ratio: str = DEFAULT_ASPECT_RATIO,
) -> Dict[str, Any]:
"""Build a uniform error response dict."""
return {
"success": False,
"image": None,
"error": error,
"error_type": error_type,
"model": model,
"prompt": prompt,
"aspect_ratio": aspect_ratio,
"provider": provider,
}