vultr/openclaw
1
0
Fork 0
mirror of https://github.com/openclaw/openclaw.git synced 2026-05-06 07:30:43 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs(media-understanding): rewrite with Steps for behavior and auto-detect, Tabs for config examples and entries, ParamField for attachments

Browse Source
This commit is contained in:
Vincent Koc
2026-04-26 00:58:16 -07:00
parent 626313a397
commit 6360e1146f
1 changed files with 294 additions and 274 deletions
Download Patch File Download Diff File Expand all files Collapse all files

568
docs/nodes/media-understanding.md
View File

@@ -4,51 +4,65 @@ read_when:
- Designing or refactoring media understanding
- Tuning inbound audio/video/image preprocessing
title: "Media understanding"
sidebarTitle: "Media understanding"
---
# Media Understanding - Inbound (2026-01-17)
OpenClaw can **summarize inbound media** (image/audio/video) before the reply pipeline runs. It auto-detects when local tools or provider keys are available, and can be disabled or customized. If understanding is off, models still receive the original files/URLs as usual.
OpenClaw can **summarize inbound media** (image/audio/video) before the reply pipeline runs. It auto‑detects when local tools or provider keys are available, and can be disabled or customized. If understanding is off, models still receive the original files/URLs as usual.
Vendor-specific media behavior is registered by vendor plugins, while OpenClaw
core owns the shared `tools.media` config, fallback order, and reply-pipeline
integration.
Vendor-specific media behavior is registered by vendor plugins, while OpenClaw core owns the shared `tools.media` config, fallback order, and reply-pipeline integration.
## Goals
- Optional: pre‑digest inbound media into short text for faster routing + better command parsing.
- Optional: pre-digest inbound media into short text for faster routing + better command parsing.
- Preserve original media delivery to the model (always).
- Support **provider APIs** and **CLI fallbacks**.
- Allow multiple models with ordered fallback (error/size/timeout).
## High-level behavior
1. Collect inbound attachments (`MediaPaths`, `MediaUrls`, `MediaTypes`).
2. For each enabled capability (image/audio/video), select attachments per policy (default: **first**).
3. Choose the first eligible model entry (size + capability + auth).
4. If a model fails or the media is too large, **fall back to the next entry**.
5. On success:
- `Body` becomes `[Image]`, `[Audio]`, or `[Video]` block.
- Audio sets `{{Transcript}}`; command parsing uses caption text when present,
otherwise the transcript.
- Captions are preserved as `User text:` inside the block.
<Steps>
<Step title="Collect attachments">
Collect inbound attachments (`MediaPaths`, `MediaUrls`, `MediaTypes`).
</Step>
<Step title="Select per-capability">
For each enabled capability (image/audio/video), select attachments per policy (default: **first**).
</Step>
<Step title="Choose model">
Choose the first eligible model entry (size + capability + auth).
</Step>
<Step title="Fallback on failure">
If a model fails or the media is too large, **fall back to the next entry**.
</Step>
<Step title="Apply success block">
On success:
- `Body` becomes `[Image]`, `[Audio]`, or `[Video]` block.
- Audio sets `{{Transcript}}`; command parsing uses caption text when present, otherwise the transcript.
- Captions are preserved as `User text:` inside the block.
</Step>
</Steps>
If understanding fails or is disabled, **the reply flow continues** with the original body + attachments.
## Config overview
`tools.media` supports **shared models** plus per‑capability overrides:
`tools.media` supports **shared models** plus per-capability overrides:
- `tools.media.models`: shared model list (use `capabilities` to gate).
- `tools.media.image` / `tools.media.audio` / `tools.media.video`:
- defaults (`prompt`, `maxChars`, `maxBytes`, `timeoutSeconds`, `language`)
- provider overrides (`baseUrl`, `headers`, `providerOptions`)
- Deepgram audio options via `tools.media.audio.providerOptions.deepgram`
- audio transcript echo controls (`echoTranscript`, default `false`; `echoFormat`)
- optional **per‑capability `models` list** (preferred before shared models)
- `attachments` policy (`mode`, `maxAttachments`, `prefer`)
- `scope` (optional gating by channel/chatType/session key)
- `tools.media.concurrency`: max concurrent capability runs (default **2**).
<AccordionGroup>
<Accordion title="Top-level keys">
- `tools.media.models`: shared model list (use `capabilities` to gate).
- `tools.media.image` / `tools.media.audio` / `tools.media.video`:
- defaults (`prompt`, `maxChars`, `maxBytes`, `timeoutSeconds`, `language`)
- provider overrides (`baseUrl`, `headers`, `providerOptions`)
- Deepgram audio options via `tools.media.audio.providerOptions.deepgram`
- audio transcript echo controls (`echoTranscript`, default `false`; `echoFormat`)
- optional **per-capability `models` list** (preferred before shared models)
- `attachments` policy (`mode`, `maxAttachments`, `prefer`)
- `scope` (optional gating by channel/chatType/session key)
- `tools.media.concurrency`: max concurrent capability runs (default **2**).
</Accordion>
</AccordionGroup>
```json5
{
@@ -77,99 +91,110 @@ If understanding fails or is disabled, **the reply flow continues** with the ori
Each `models[]` entry can be **provider** or **CLI**:
```json5
{
type: "provider", // default if omitted
provider: "openai",
model: "gpt-5.5",
prompt: "Describe the image in <= 500 chars.",
maxChars: 500,
maxBytes: 10485760,
timeoutSeconds: 60,
capabilities: ["image"], // optional, used for multi‑modal entries
profile: "vision-profile",
preferredProfile: "vision-fallback",
}
```
<Tabs>
<Tab title="Provider entry">
```json5
{
type: "provider", // default if omitted
provider: "openai",
model: "gpt-5.5",
prompt: "Describe the image in <= 500 chars.",
maxChars: 500,
maxBytes: 10485760,
timeoutSeconds: 60,
capabilities: ["image"], // optional, used for multi-modal entries
profile: "vision-profile",
preferredProfile: "vision-fallback",
}
```
</Tab>
<Tab title="CLI entry">
```json5
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
],
maxChars: 500,
maxBytes: 52428800,
timeoutSeconds: 120,
capabilities: ["video", "image"],
}
```
```json5
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
],
maxChars: 500,
maxBytes: 52428800,
timeoutSeconds: 120,
capabilities: ["video", "image"],
}
```
CLI templates can also use:
CLI templates can also use:
- `{{MediaDir}}` (directory containing the media file)
- `{{OutputDir}}` (scratch dir created for this run)
- `{{OutputBase}}` (scratch file base path, no extension)
- `{{MediaDir}}` (directory containing the media file)
- `{{OutputDir}}` (scratch dir created for this run)
- `{{OutputBase}}` (scratch file base path, no extension)
</Tab>
</Tabs>
## Defaults and limits
Recommended defaults:
- `maxChars`: **500** for image/video (short, command‑friendly)
- `maxChars`: **500** for image/video (short, command-friendly)
- `maxChars`: **unset** for audio (full transcript unless you set a limit)
- `maxBytes`:
- image: **10MB**
- audio: **20MB**
- video: **50MB**
Rules:
- If media exceeds `maxBytes`, that model is skipped and the **next model is tried**.
- Audio files smaller than **1024 bytes** are treated as empty/corrupt and skipped before provider/CLI transcription; inbound reply context receives a deterministic placeholder transcript so the agent knows the note was too small.
- If the model returns more than `maxChars`, output is trimmed.
- `prompt` defaults to simple “Describe the {media}.” plus the `maxChars` guidance (image/video only).
- If the active primary image model already supports vision natively, OpenClaw
skips the `[Image]` summary block and passes the original image into the
model instead.
- If a Gateway/WebChat primary model is text-only, image attachments are
preserved as offloaded `media://inbound/*` refs so the image/PDF tools or
configured image model can still inspect them instead of losing the attachment.
- Explicit `openclaw infer image describe --model <provider/model>` requests
are different: they run that image-capable provider/model directly, including
Ollama refs such as `ollama/qwen2.5vl:7b`.
- If `<capability>.enabled: true` but no models are configured, OpenClaw tries the
**active reply model** when its provider supports the capability.
<AccordionGroup>
<Accordion title="Rules">
- If media exceeds `maxBytes`, that model is skipped and the **next model is tried**.
- Audio files smaller than **1024 bytes** are treated as empty/corrupt and skipped before provider/CLI transcription; inbound reply context receives a deterministic placeholder transcript so the agent knows the note was too small.
- If the model returns more than `maxChars`, output is trimmed.
- `prompt` defaults to simple "Describe the {media}." plus the `maxChars` guidance (image/video only).
- If the active primary image model already supports vision natively, OpenClaw skips the `[Image]` summary block and passes the original image into the model instead.
- If a Gateway/WebChat primary model is text-only, image attachments are preserved as offloaded `media://inbound/*` refs so the image/PDF tools or configured image model can still inspect them instead of losing the attachment.
- Explicit `openclaw infer image describe --model <provider/model>` requests are different: they run that image-capable provider/model directly, including Ollama refs such as `ollama/qwen2.5vl:7b`.
- If `<capability>.enabled: true` but no models are configured, OpenClaw tries the **active reply model** when its provider supports the capability.
</Accordion>
</AccordionGroup>
### Auto-detect media understanding (default)
If `tools.media.<capability>.enabled` is **not** set to `false` and you haven’t
configured models, OpenClaw auto-detects in this order and **stops at the first
working option**:
If `tools.media.<capability>.enabled` is **not** set to `false` and you haven't configured models, OpenClaw auto-detects in this order and **stops at the first working option**:
1. **Active reply model** when its provider supports the capability.
2. **`agents.defaults.imageModel`** primary/fallback refs (image only).
3. **Local CLIs** (audio only; if installed)
- `sherpa-onnx-offline` (requires `SHERPA_ONNX_MODEL_DIR` with encoder/decoder/joiner/tokens)
- `whisper-cli` (`whisper-cpp`; uses `WHISPER_CPP_MODEL` or the bundled tiny model)
- `whisper` (Python CLI; downloads models automatically)
4. **Gemini CLI** (`gemini`) using `read_many_files`
5. **Provider auth**
- Configured `models.providers.*` entries that support the capability are
tried before the bundled fallback order.
- Image-only config providers with an image-capable model auto-register for
media understanding even when they are not a bundled vendor plugin.
- Ollama image understanding is available when selected explicitly, for
example through `agents.defaults.imageModel` or
`openclaw infer image describe --model ollama/<vision-model>`.
- Bundled fallback order:
- Audio: OpenAI → Groq → xAI → Deepgram → Google → SenseAudio → ElevenLabs → Mistral
- Image: OpenAI → Anthropic → Google → MiniMax → MiniMax Portal → Z.AI
- Video: Google → Qwen → Moonshot
<Steps>
<Step title="Active reply model">
Active reply model when its provider supports the capability.
</Step>
<Step title="agents.defaults.imageModel">
`agents.defaults.imageModel` primary/fallback refs (image only).
</Step>
<Step title="Local CLIs (audio only)">
Local CLIs (if installed):
- `sherpa-onnx-offline` (requires `SHERPA_ONNX_MODEL_DIR` with encoder/decoder/joiner/tokens)
- `whisper-cli` (`whisper-cpp`; uses `WHISPER_CPP_MODEL` or the bundled tiny model)
- `whisper` (Python CLI; downloads models automatically)
</Step>
<Step title="Gemini CLI">
`gemini` using `read_many_files`.
</Step>
<Step title="Provider auth">
- Configured `models.providers.*` entries that support the capability are tried before the bundled fallback order.
- Image-only config providers with an image-capable model auto-register for media understanding even when they are not a bundled vendor plugin.
- Ollama image understanding is available when selected explicitly, for example through `agents.defaults.imageModel` or `openclaw infer image describe --model ollama/<vision-model>`.
Bundled fallback order:
- Audio: OpenAI → Groq → xAI → Deepgram → Google → SenseAudio → ElevenLabs → Mistral
- Image: OpenAI → Anthropic → Google → MiniMax → MiniMax Portal → Z.AI
- Video: Google → Qwen → Moonshot
</Step>
</Steps>
To disable auto-detection, set:
@@ -185,26 +210,24 @@ To disable auto-detection, set:
}
```
Note: Binary detection is best-effort across macOS/Linux/Windows; ensure the CLI is on `PATH` (we expand `~`), or set an explicit CLI model with a full command path.
<Note>
Binary detection is best-effort across macOS/Linux/Windows; ensure the CLI is on `PATH` (we expand `~`), or set an explicit CLI model with a full command path.
</Note>
### Proxy environment support (provider models)
When provider-based **audio** and **video** media understanding is enabled, OpenClaw
honors standard outbound proxy environment variables for provider HTTP calls:
When provider-based **audio** and **video** media understanding is enabled, OpenClaw honors standard outbound proxy environment variables for provider HTTP calls:
- `HTTPS_PROXY`
- `HTTP_PROXY`
- `https_proxy`
- `http_proxy`
If no proxy env vars are set, media understanding uses direct egress.
If the proxy value is malformed, OpenClaw logs a warning and falls back to direct
fetch.
If no proxy env vars are set, media understanding uses direct egress. If the proxy value is malformed, OpenClaw logs a warning and falls back to direct fetch.
## Capabilities (optional)
If you set `capabilities`, the entry only runs for those media types. For shared
lists, OpenClaw can infer defaults:
If you set `capabilities`, the entry only runs for those media types. For shared lists, OpenClaw can infer defaults:
- `openai`, `anthropic`, `minimax`: **image**
- `minimax-portal`: **image**
@@ -217,11 +240,9 @@ lists, OpenClaw can infer defaults:
- `groq`: **audio**
- `xai`: **audio**
- `deepgram`: **audio**
- Any `models.providers.<id>.models[]` catalog with an image-capable model:
**image**
- Any `models.providers.<id>.models[]` catalog with an image-capable model: **image**
For CLI entries, **set `capabilities` explicitly** to avoid surprising matches.
If you omit `capabilities`, the entry is eligible for the list it appears in.
For CLI entries, **set `capabilities` explicitly** to avoid surprising matches. If you omit `capabilities`, the entry is eligible for the list it appears in.
## Provider support matrix (OpenClaw integrations)
@@ -231,12 +252,12 @@ If you omit `capabilities`, the entry is eligible for the list it appears in.
| Audio | OpenAI, Groq, xAI, Deepgram, Google, SenseAudio, ElevenLabs, Mistral | Provider transcription (Whisper/Groq/xAI/Deepgram/Gemini/SenseAudio/Scribe/Voxtral). |
| Video | Google, Qwen, Moonshot | Provider video understanding via vendor plugins; Qwen video understanding uses the Standard DashScope endpoints. |
MiniMax note:
<Note>
**MiniMax note**
- `minimax` and `minimax-portal` image understanding comes from the plugin-owned
`MiniMax-VL-01` media provider.
- The bundled MiniMax text catalog still starts text-only; explicit
`models.providers.minimax` entries materialize image-capable M2.7 chat refs.
- `minimax` and `minimax-portal` image understanding comes from the plugin-owned `MiniMax-VL-01` media provider.
- The bundled MiniMax text catalog still starts text-only; explicit `models.providers.minimax` entries materialize image-capable M2.7 chat refs.
</Note>
## Model selection guidance
@@ -248,177 +269,176 @@ MiniMax note:
## Attachment policy
Per‑capability `attachments` controls which attachments are processed:
Per-capability `attachments` controls which attachments are processed:
- `mode`: `first` (default) or `all`
- `maxAttachments`: cap the number processed (default **1**)
- `prefer`: `first`, `last`, `path`, `url`
<ParamField path="mode" type='"first" | "all"' default="first">
Whether to process the first selected attachment or all of them.
</ParamField>
<ParamField path="maxAttachments" type="number" default="1">
Cap the number processed.
</ParamField>
<ParamField path="prefer" type='"first" | "last" | "path" | "url"'>
Selection preference among candidate attachments.
</ParamField>
When `mode: "all"`, outputs are labeled `[Image 1/2]`, `[Audio 2/2]`, etc.
File-attachment extraction behavior:
- Extracted file text is wrapped as **untrusted external content** before it is
appended to the media prompt.
- The injected block uses explicit boundary markers like
`<<<EXTERNAL_UNTRUSTED_CONTENT id="...">>>` /
`<<<END_EXTERNAL_UNTRUSTED_CONTENT id="...">>>` and includes a
`Source: External` metadata line.
- This attachment-extraction path intentionally omits the long
`SECURITY NOTICE:` banner to avoid bloating the media prompt; the boundary
markers and metadata still remain.
- If a file has no extractable text, OpenClaw injects `[No extractable text]`.
- If a PDF falls back to rendered page images in this path, the media prompt keeps
the placeholder `[PDF content rendered to images; images not forwarded to model]`
because this attachment-extraction step forwards text blocks, not the rendered PDF images.
<AccordionGroup>
<Accordion title="File-attachment extraction behavior">
- Extracted file text is wrapped as **untrusted external content** before it is appended to the media prompt.
- The injected block uses explicit boundary markers like `<<<EXTERNAL_UNTRUSTED_CONTENT id="...">>>` / `<<<END_EXTERNAL_UNTRUSTED_CONTENT id="...">>>` and includes a `Source: External` metadata line.
- This attachment-extraction path intentionally omits the long `SECURITY NOTICE:` banner to avoid bloating the media prompt; the boundary markers and metadata still remain.
- If a file has no extractable text, OpenClaw injects `[No extractable text]`.
- If a PDF falls back to rendered page images in this path, the media prompt keeps the placeholder `[PDF content rendered to images; images not forwarded to model]` because this attachment-extraction step forwards text blocks, not the rendered PDF images.
</Accordion>
</AccordionGroup>
## Config examples
### 1) Shared models list + overrides
```json5
{
tools: {
media: {
models: [
{ provider: "openai", model: "gpt-5.5", capabilities: ["image"] },
{
provider: "google",
model: "gemini-3-flash-preview",
capabilities: ["image", "audio", "video"],
},
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
<Tabs>
<Tab title="Shared models + overrides">
```json5
{
tools: {
media: {
models: [
{ provider: "openai", model: "gpt-5.5", capabilities: ["image"] },
{
provider: "google",
model: "gemini-3-flash-preview",
capabilities: ["image", "audio", "video"],
},
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
],
capabilities: ["image", "video"],
},
],
capabilities: ["image", "video"],
audio: {
attachments: { mode: "all", maxAttachments: 2 },
},
video: {
maxChars: 500,
},
},
],
audio: {
attachments: { mode: "all", maxAttachments: 2 },
},
video: {
maxChars: 500,
},
},
},
}
```
### 2) Audio + Video only (image off)
```json5
{
tools: {
media: {
audio: {
enabled: true,
models: [
{ provider: "openai", model: "gpt-4o-mini-transcribe" },
{
type: "cli",
command: "whisper",
args: ["--model", "base", "{{MediaPath}}"],
},
],
},
video: {
enabled: true,
maxChars: 500,
models: [
{ provider: "google", model: "gemini-3-flash-preview" },
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
}
```
</Tab>
<Tab title="Audio + video only">
```json5
{
tools: {
media: {
audio: {
enabled: true,
models: [
{ provider: "openai", model: "gpt-4o-mini-transcribe" },
{
type: "cli",
command: "whisper",
args: ["--model", "base", "{{MediaPath}}"],
},
],
},
],
},
},
},
}
```
### 3) Optional image understanding
```json5
{
tools: {
media: {
image: {
enabled: true,
maxBytes: 10485760,
maxChars: 500,
models: [
{ provider: "openai", model: "gpt-5.5" },
{ provider: "anthropic", model: "claude-opus-4-6" },
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
video: {
enabled: true,
maxChars: 500,
models: [
{ provider: "google", model: "gemini-3-flash-preview" },
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
],
},
],
},
],
},
},
},
},
}
```
### 4) Multi-modal single entry (explicit capabilities)
```json5
{
tools: {
media: {
image: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
}
```
</Tab>
<Tab title="Image-only">
```json5
{
tools: {
media: {
image: {
enabled: true,
maxBytes: 10485760,
maxChars: 500,
models: [
{ provider: "openai", model: "gpt-5.5" },
{ provider: "anthropic", model: "claude-opus-4-6" },
{
type: "cli",
command: "gemini",
args: [
"-m",
"gemini-3-flash",
"--allowed-tools",
"read_file",
"Read the media at {{MediaPath}} and describe it in <= {{MaxChars}} characters.",
],
},
],
},
],
},
},
audio: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
}
```
</Tab>
<Tab title="Multi-modal single entry">
```json5
{
tools: {
media: {
image: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
},
],
},
],
},
video: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
audio: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
},
],
},
],
video: {
models: [
{
provider: "google",
model: "gemini-3.1-pro-preview",
capabilities: ["image", "video", "audio"],
},
],
},
},
},
},
},
}
```
}
```
</Tab>
</Tabs>
## Status output
@@ -428,15 +448,15 @@ When media understanding runs, `/status` includes a short summary line:
📎 Media: image ok (openai/gpt-5.4) · audio skipped (maxBytes)
```
This shows per‑capability outcomes and the chosen provider/model when applicable.
This shows per-capability outcomes and the chosen provider/model when applicable.
## Notes
- Understanding is **best‑effort**. Errors do not block replies.
- Understanding is **best-effort**. Errors do not block replies.
- Attachments are still passed to models even when understanding is disabled.
- Use `scope` to limit where understanding runs (e.g. only DMs).
## Related docs
## Related
- [Configuration](/gateway/configuration)
- [Image & Media Support](/nodes/images)
- [Image & media support](/nodes/images)