SLIDESHOT [docs] Markdown
For agents: read the markdown version of these docs →

Feedback

The feedback endpoint sends a freeform message to the Slideshot team, optionally tied to one or more runs the caller has access to. Use it to report quality issues, surprising agent behavior, or simply to share when a run produced an exceptional result.

When to send feedback

  • A run failed in an unexpected way.
  • The polished demo did not match the workflow you described.
  • The agent picked confusing or wrong run options.
  • You want to flag something positive worth keeping in future runs.

Agents using the MCP server or the skill should prefer sending feedback automatically when a run fails or the user explicitly asks them to.

Endpoint

  • POST /v1/agent/feedback
    • Body: { "feedback": "...", "run_ids": ["<uuid>"] }
    • feedback is required plain text. Trimmed, max 10000 characters.
    • run_ids is optional. When provided, every run must belong to the calling API key. Deduplicated, max 20 entries.
    • Response: { "created_at": "<iso>" }.

Example

curl -X POST https://api.slideshot.ai/v1/agent/feedback \
  -H "x-api-key: $SLIDESHOT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "feedback": "The intro card skipped on the last two runs even though I set a template. The demo still shipped without it.",
    "run_ids": [
      "11111111-2222-3333-4444-555555555555",
      "66666666-7777-8888-9999-aaaabbbbcccc"
    ]
  }'
{ "created_at": "2026-05-11T12:10:00Z" }

Error cases

  • 400: Payload is invalid, or the account has no reply-to email configured.
  • 404: One of the referenced runs is missing or not visible to the calling API key.
  • 500: The submission was stored but the reply email could not be sent. The feedback still reaches the team; treat this as a delivery-only failure and avoid retrying the same body.

Tips for useful feedback

  • Include run_ids whenever you can. It lets the team look at the exact run that produced the issue.
  • Describe the expected outcome, not just the symptom. “Intro card was skipped” is helpful; “expected the intro card to render but demo.mp4 opens directly with the workflow” is more helpful.
  • Mention the visual options you set, especially for video.background, video.size, and intro.
  • For successful runs, name what you liked. That helps the defaults the agent picks next time.
  • Runs: the runs you reference in run_ids.
  • Artifacts: what landed in the run you are giving feedback on.
  • API overview: request shape and error model.