tdfs4ds.agentops.client¶
Thin REST client for the AgentOps Core API.
Deliberately no orchestration: one method per endpoint, so the pipeline in
:mod:tdfs4ds.agentops.deploy can be tested against a stub without a cluster.
Three details cost a real deployment cycle each to discover, and are encoded here rather than left to the caller:
- the multipart upload must not set
Content-Type—requestsderives the boundary, and supplying the header without one produces a request the API cannot parse; - the payload is rejected outright for any field the schema does not define
(
422 extra_forbidden), so nothing decorative may be added to it; - deployment progress is reported as
status+current_step+steps_remaining, which is what to poll — a fixed sleep either returns before the pod is up or wastes minutes after it is.
The token is used and never logged, echoed, or returned.
module-attribute
¶
module-attribute
¶
Bases: RuntimeError
An AgentOps API call failed. Carries the response body, which is where the useful detail lives — a 422 names the offending field.
Minimal client for the endpoints a deployment needs.
Parameters¶
base_url : str, optional
Defaults to TD_AGENTOPS_BASE_URL, then the in-cluster service URL.
token : str, optional
JWT bearer token. Defaults to TDFS4DS_AGENTOPS_TOKEN. In an AI
Studio notebook this is the user_token produced by %gen_token.
Upload the zip and register an agent definition. Returns its id.
Poll until the definition leaves creating. Returns its status.
POST the deployment payload. Returns the full response body.
The payload is sent verbatim; see the module docstring on
extra_forbidden.
Poll the deployment until it settles. Returns the last body.
Emits current_step transitions rather than every poll, so a long
build reads as building -> pushing -> deploying instead of forty
identical lines.
Pod logs. Never raises — a missing log must not mask the real error.