You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/nodes/media-understanding.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -321,7 +321,7 @@ When `mode: "all"`, outputs are labeled `[Image 1/2]`, `[Audio 2/2]`, etc.
321
321
- The injected block uses explicit boundary markers like `<<<EXTERNAL_UNTRUSTED_CONTENT id="...">>>` / `<<<END_EXTERNAL_UNTRUSTED_CONTENT id="...">>>` and includes a `Source: External` metadata line.
322
322
- This attachment-extraction path intentionally omits the long `SECURITY NOTICE:` banner to avoid bloating the media prompt; the boundary markers and metadata still remain.
323
323
- If a file has no extractable text, OpenClaw injects `[No extractable text]`.
324
-
- 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.
324
+
- If a PDF falls back to rendered page images in this path, OpenClaw forwards those page images to vision-capable reply models and keeps the placeholder `[PDF content rendered to images]` in the file block.
@@ -132,15 +195,20 @@ export async function resolveCurrentTurnImages(params: {
132
195
logVerbose(
133
196
`agent-runner: native OpenClaw media resolution produced ${images.length}/${undescribedImageAttachments.length} current image attachment(s); falling back to prompt image refs`,
0 commit comments