brand_feedback
Send feedback signals to the brandsystem team about tool errors, friction, missing features, data quality issues, or positive experiences to drive workflow improvements.
Instructions
Send a feedback signal to the brandsystem team about a tool or workflow. Use when a tool errors, extraction misses obvious data, the workflow felt harder than it should, an agent's path got blocked, or something worked particularly well and should be preserved. Set category to one of: 'bug' (broken), 'friction' (works but painful), 'feature_request' (capability missing), 'data_quality' (results wrong/incomplete), 'praise' (worth keeping), 'agent_signal' (structured telemetry — also pass signal + tool_used + signal_context; brand context auto-fills from .brand/config). Provide a one-line summary plus optional detail/severity/context. Writes to ~/.brandsystem/feedback/ AND attempts a fire-and-forget POST to brandcode.studio (drains backlog if previously offline). Returns the feedback ID and remote-send status. NOT for reading existing feedback — use brand_feedback_review. NOT for changing item status — use brand_feedback_triage.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| detail | No | Full context: what the agent was trying to do, what happened, what was expected, and any suggested fix. Use this for the complete feedback body — up to 10,000 characters. | |
| signal | No | Signal type. Required when category is 'agent_signal'. 'positive': tool worked well, 'negative': tool failed or gave poor results, 'suggestion': improvement idea. | |
| context | No | Optional structured context about the session. | |
| message | No | Alias for 'detail'. Full feedback body — what happened, what was expected, reproduction steps, suggested fix. Either 'message' or 'detail' can be used; if both provided, they are concatenated. | |
| outcome | No | What happened as a result. Optional for positive signals. | |
| summary | Yes | One-line summary of the feedback. | |
| category | Yes | Type of feedback. 'bug': something is broken. 'friction': it works but is harder than it should be. 'feature_request': a tool or capability that should exist. 'data_quality': extraction results seem wrong or incomplete. 'praise': something that works well and should be preserved. 'agent_signal': structured signal from an agent about tool usage (requires signal, tool_used, signal_context). | |
| severity | No | How much this impacts the agent's ability to serve the user. Defaults to 'suggestion'. | |
| tool_name | No | Which brandsystem tool this feedback relates to (e.g. 'brand_extract_web', 'brand_compile'). Optional for general feedback. | |
| tool_used | No | Which tool triggered this signal (e.g. 'brand_extract_web'). Required when category is 'agent_signal'. | |
| signal_context | No | What the agent was trying to do when this signal occurred. Required when category is 'agent_signal'. |