Prompt Logs

Query LLM prompt logs for auditing and debugging agent conversations across a workspace.

The Prompt Logs endpoint lets you retrieve the full input and output of every LLM call made during agent conversations - system prompts, conversation history, tool catalogs, model selection, and responses. This is useful for auditing agent behavior, debugging unexpected responses, and understanding how context flows through multi-turn conversations.

Access restriction: Prompt log access requires an API key with the admin or owner role. Prompt logs contain full conversation history and tool I/O, which may include PHI. All reads are audit-logged.

List Prompt Logs

GET /v1/{workspace_id}/prompt-logs

Returns prompt log entries for the workspace, newest first. Typical response latency is 1-5 seconds.

Query Parameters

Parameter	Type	Required	Description
call_sid	string	No	Filter to a single call (max 64 characters)
prompt_type	string	No	Filter by prompt type, e.g. engage_user, navigation, tool (max 64 characters)
state_name	string	No	Filter by context graph state name at render time (max 128 characters)
from_ts	datetime	No	Inclusive lower bound on event timestamp
to_ts	datetime	No	Exclusive upper bound on event timestamp
limit	integer	No	Max rows to return (default 20, min 1, max 200)
offset	integer	No	Offset for pagination (default 0, max 10000)

Auto-applied time window: When no call_sid, from_ts, or to_ts filter is provided, the query is automatically capped to the last 7 days. The applied window is reported in the applied_time_window_days response field.

Response

Field	Type	Description
items	array	List of prompt log entries
count	integer	Number of entries in this page
has_more	boolean	Whether more rows exist beyond this page
next_offset	integer or null	Pass back as offset to fetch the next page. Null when has_more is false
applied_time_window_days	integer or null	Auto-applied time bound (in days) when no explicit time or call filter was provided. Null when the caller supplied a selectivity-bearing filter

Prompt Log Entry

Field	Type	Description
event_id	string	Unique event identifier
call_sid	string or null	Call or session identifier
workspace_id	string	Workspace identifier
effective_at	datetime or null	When the event occurred
ingested_at	datetime or null	When the event was ingested
source	string or null	Event source
source_system	string or null	Originating system
prompt_type	string or null	Type of prompt (e.g. engage_user, navigation, tool)
turn_index	integer or null	Turn number within the conversation
system_prompt	string or null	Full system prompt sent to the model
history	array or null	Conversation history as a list of role/content message objects
full_prompt	string or null	Complete assembled prompt
llm_model	string or null	Model identifier used for this call
llm_response	string or null	Raw model response
state_name	string or null	Context graph state at render time
action	string or null	Action being performed
tool_names	array	List of tool names available for this call
has_tools	boolean or null	Whether tools were available
service_id	string or null	Service identifier
session_id	string or null	Session identifier
data_parse_error	boolean	True when the underlying event data could not be fully parsed. Populated fields fall back to defaults

String fields like system_prompt, full_prompt, and llm_response reflect the actual LLM input/output and can be large (typically 1-50 KB each, occasionally larger for long conversation histories). Do not assume a max length.

Pagination

The endpoint uses peek-ahead pagination. When has_more is true, pass the value of next_offset as the offset parameter on the next request to retrieve the next page.

Error Responses

Status	Description
403 Forbidden	Caller does not have admin or owner role
503 Service Unavailable	Prompt log data source is unavailable

Example

curl -X GET \
  'https://api.platform.amigo.ai/v1/{workspace_id}/prompt-logs?call_sid=CA1234567890&limit=10' \
  -H 'Authorization: Bearer {api_key}'

{
  "items": [
    {
      "event_id": "evt_abc123",
      "call_sid": "CA1234567890",
      "workspace_id": "ws_xyz",
      "effective_at": "2026-07-15T14:30:00Z",
      "prompt_type": "engage_user",
      "turn_index": 3,
      "system_prompt": "You are a helpful healthcare scheduling assistant...",
      "history": [
        {"role": "user", "content": "I need to reschedule my appointment"},
        {"role": "assistant", "content": "I can help with that..."}
      ],
      "llm_model": "model-v2",
      "llm_response": "Let me look up your upcoming appointments...",
      "state_name": "scheduling",
      "tool_names": ["search_appointments", "reschedule_appointment"],
      "has_tools": true,
      "data_parse_error": false
    }
  ],
  "count": 1,
  "has_more": false,
  "next_offset": null,
  "applied_time_window_days": null
}