Vito Sansevero
25cbf43030
Record the shared video carrier order (per-key tags, comment envelope, description), the fold-duplicate rule, the 6 MiB + 64 KiB probe bound with the role-only re-probe and limited workflow, and that scans read the same carrier as the inspector.
2.3 KiB
Prompt inspection and sending
The opt-in workspace at /prompt_manager/workspace#/images shows full positive
and negative prompts in both Image details and the full-image preview. Existing
gallery and metadata tools remain available during the application rebuild.
Each prompt shows its evidence source: recorded execution, workflow inference, or embedded generation parameters. Path-only saved-library associations are marked unverified and cannot be sent: a filename alone cannot prove that an image has not been replaced. Missing, ambiguous, and unreadable prompts remain visible as distinct states. An explicitly empty negative is a resolved value and can clear a destination's negative field.
Use the blue paper-plane action beside a prompt, or Send prompt pair. Keep a ComfyUI tab open under the same origin and deployment prefix. Choose a window and compatible Prompt Manager node, select Preview changes, review the complete before/after text and preserved prepend/append composition, then Apply to ComfyUI. The dialog reports success only after a matching acknowledgement. It does not load workflows or queue generation. If confirmation is lost, inspect ComfyUI and preview again before another attempt. Closing the dialog cannot undo an update already sent.
Read-only API
GET /prompt_manager/images/inspect/{relative_path}?root={root_index} requires
an explicit configured root, including when two roots contain the same filename.
The response contains image identity and prompts.positive / prompts.negative.
Each role has status (resolved, absent, unresolved, or ambiguous), text
(string only when resolved, otherwise null), and source. Inspection does not
index files, alter saved prompts, run scans, or decode image pixels. PNG text and
video probe output are bounded; unsupported or malformed evidence remains
unresolved rather than becoming sendable placeholder text. Videos are read from
one container carrier: the per-key tags written by core SaveVideo/SaveWEBM and
current VideoHelperSuite (prompt, workflow, parameters,
comfyassets_prompt_pairs), else the old VideoHelperSuite comment envelope,
else description. The scan reads the same carrier, so Images and ingestion
agree on a video's prompts (see TESTING.md for the order and budgets).
Python route changes require restarting ComfyUI. Browser assets require a refresh.