Skip to main content
GET
Python SDK

Authorizations

​
Authorization
string
header
required

Enter your API key (e.g. dai_sk_test_key_001)

Headers

​
Authorization
string
​
X-Workspace-Id
string | null

Query Parameters

​
limit
integer
default:20
Required range: 1 <= x <= 100
​
offset
integer
default:0
Required range: x >= 0
​
status
enum<string> | null

Filter by trace status. Allowed: success/error/degraded.

Available options:
success,
error,
degraded
​
agent_name
string | null
​
eval_verdict
string | null

Filter by eval verdict (pass/fail/keep/drop/review/...), or 'unevaluated' for traces with no verdict yet.

​
source_type
enum<string> | null

Filter by trace source_type. See ingest allowlist for valid values.

Available options:
production,
test,
evaluation,
sdk,
manual,
synthetic,
development,
sandbox,
distillation,
snapshot,
file,
url,
eval_replay,
sample,
demo
​
manifest_id
string | null
​
search
string | null

Full-text search across user_input, final_output, and agent_name

​
include_fixtures
boolean
default:true

When false, drop platform-harness fixture agents (__, live-, cuj*, soak-*, …) from the listing — the UI's default. Defaults to true for SDK/API back-compat. Ignored when an explicit agent/skill/tool filter is set (asking by name always works).

​
target_manifest_id
string | null

Show compatibility of traces relative to this manifest

​
child_agent_name
string | null

Filter parent traces by child agent name (multi-agent)

​
child_status
enum<string> | null

Filter child agent matches by status. Allowed: success/error/degraded (H1091).

Available options:
success,
error,
degraded
​
tool_name
string | null

Filter to traces that invoked the named tool (uses ToolCallIndex)

​
skill_name
string | null

Filter to traces where the named skill was activated (uses trace_skill_activation). Backs the owner skill-detail live-traces panel (improve#1413).

​
has_spans
boolean | null

H1336 (improve#702): when true, exclude traces with 0 spans (filters out empty SDK / Hello-World traces that pollute investigation views — caught by cuj07 security-sleuth at #701). When false, only return empty-span traces. Omit for both.

​
has_llm_calls
boolean | null

H1336 (improve#702): symmetric filter for traces with 0 LLM calls.

​
sort
enum<string> | null

H1092 (improve#843): order results by the given field. Prefix '-' for descending. Defaults to '-created_at'. Unknown values 422 instead of being silently ignored.

Available options:
-created_at,
created_at,
-started_at,
started_at
​
started_at_from
string<date-time> | null

H1093 (improve#863): inclusive lower bound on RunTrace.started_at (ISO 8601). Malformed dates 422 via Pydantic typing instead of being silently ignored.

​
started_at_to
string<date-time> | null

H1093 (improve#863): inclusive upper bound on RunTrace.started_at (ISO 8601). Same validation behavior as started_at_from.

​
compat_status
string | null

Filter to traces with this compatibility verdict against target_manifest_id. Values: keep / repair / replay / drop. Requires target_manifest_id to be set for authoritative scoping; otherwise falls back to RunTrace.compat_status (a less-precise denormalized field).

Cookies

​
decimal_session
string | null

Response

Successful Response

The response is of type Response List Traces Api V1 Traces Get · object.