datarobot-workload-api

Name: datarobot-workload-api
Author: datarobot-oss

22stars19forks

How to install datarobot-workload-api?

datarobot-oss/datarobot-agent-skills/datarobot-workload-api

$npx -y skills add datarobot-oss/datarobot-agent-skills --skill datarobot-workload-api

Installs into the current project.

Summary

Use when the user wants to create, configure, scale, debug, observe, or roll out container workloads on DataRobot's Workload API. Triggers include: deploying a container as a managed service, listing/starting/stopping workloads, changing replica counts or autoscaling, picking CPU

Files · 1

View on GitHub

SKILL.md

1	# DataRobot Workload API
2
3	Run container images as managed, autoscalable services on DataRobot. One skill, four jobs — pick the section by user intent:
4
5	1. Create / configure / scale — deploy a container; change replicas, resources, autoscaling, bundle; inject credentials
6	2. Diagnose — workload is stuck, errored, or crash-looping
7	3. Observe — logs, traces, metrics, service stats for a running workload
8	4. Artifact lifecycle — iterate drafts, build images, lock for production, roll out new versions
9
10	## Prerequisites
11
12	`DATAROBOT_ENDPOINT` (must end in `/api/v2`) and `DATAROBOT_API_TOKEN` must be set. Run `datarobot-setup` if not. Auth header: `Authorization: Bearer ${DATAROBOT_API_TOKEN}`. The Workload API is not in the `datarobot` Python SDK — call REST directly.
13
14	Transport. Examples use Python `httpx` (`pip install httpx`). The API is plain HTTP, so equivalent calls work via `curl` or the `pulumi-datarobot` Pulumi provider declaratively. The skill teaches the model; transport is interchangeable.
15
16	## Bundled scripts
17
18	Runnable Python in `scripts/` (this skill's folder). Each uses `httpx` and reads `DATAROBOT_ENDPOINT` + `DATAROBOT_API_TOKEN`:
19
20	- `wait_for_running.py <workload_id>` — poll until `running`; exit 2 on terminal failure, 3 on timeout
21	- `diagnose_workload.py <workload_id>` — run the 5-step debug flow, print a structured diagnosis (`--json` for machine-readable)
22	- `wait_for_build.py <artifact_id> <build_id>` — poll a server-side image build; dumps last 2KB of logs on `FAILED`
23	- `wait_for_replacement.py <workload_id>` — poll a rolling replacement; handles the 404-when-cleared case
24	- `check_limits.py` — print the user's effective org-set scaling limits via `/account/info/`
25
26	## Deeper docs in references/
27
28	The SKILL.md is the operational core. Detail an agent needs occasionally lives in `references/`:
29
30	- `references/status-vocabulary.md` — workload + proton status enums and lifecycle transitions
31	- `references/common-error-patterns.md` — `CrashLoopBackOff` / `ImagePullBackOff` / `OOMKilled` / probe failures / `exec format error` / pending pods
32	- `references/schema-reference.md` — OpenAPI schemas worth looking up, credential type → key mappings, public-spec path-key quirks
33	- `references/lifecycle-flows.md` — artifact draft → lock → production flow rules and behavioral gotchas
34	- `references/code-to-workload.md` — deploy from source code (no Dockerfile authoring): `dr` CLI commands, `codeRef`, Execution Environments, generated vs provided Dockerfile modes, iterate-rebuild loop
35
36	## OpenAPI spec is source of truth
37
38	Published at `https://docs.datarobot.com/en/docs/api/reference/public-api/openapi.yaml`. The spec is ~5 MB — never dump it whole into context. Save once, then extract targeted slices with `yq`:
39
40	```bash
41	curl -sS "${DATAROBOT_ENDPOINT}/openapi.yaml" -o /tmp/wapi-spec.yaml
42	yq '.components.schemas.CreateWorkloadRequest' /tmp/wapi-spec.yaml
43	yq '.paths."/api/v2/workloads/{workload_id}/".patch' /tmp/wapi-spec.yaml
44	yq '.components.schemas \| keys \| .[]' /tmp/wapi-spec.yaml \| grep -i workload # discover
45	```
46
47	Python fallback: only `print()` the specific key, never the parsed dict. All workload paths in the public spec are keyed with `/api/v2/` prefix — see `references/schema-reference.md`.
48
49	---
50
51	# 1. Create / configure / scale
52
53	## Run a container as a workload (the 90% case)
54
55	```yaml
56	# spec.yaml — JSON also accepted; spec is sent verbatim
57	name: my-api-service
58	importance: low
59	artifact:
60	name: my-api-service-artifact
61	spec:
62	type: service
63	containerGroups:
64	- name: default
65	containers:
66	- name: main
67	imageUri: ghcr.io/org/my-app:latest
68	port: 8000
69	primary: true
70	readinessProbe: {path: /readyz, port: 8000, initialDelaySeconds: 10}
71	livenessProbe: {path: /healthz, port: 8000, initialDelaySeconds: 30}
72	runtime:
73	containerGroups:
74	- name: default # must match artifact.spec.containerGroups[].name (above)
75	replicaCount: 1
76	containers:
77	- name: main
78	resourceAllocation: {cpu: 1, memory: "512MB"}
79	```
80
81	```bash
82	dr workload create --spec-file spec.yaml # v0.2.74+; 4xx: 400=schema/li

People who liked this also viewed

6 picks

Type

datarobot-workload-api

22stars19forks

View on GitHub

How to install datarobot-workload-api?

datarobot-oss/datarobot-agent-skills/datarobot-workload-api

$npx -y skills add datarobot-oss/datarobot-agent-skills --skill datarobot-workload-api

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# DataRobot Workload API
2
3	Run container images as managed, autoscalable services on DataRobot. One skill, four jobs — pick the section by user intent:
4
5	1. Create / configure / scale — deploy a container; change replicas, resources, autoscaling, bundle; inject credentials
6	2. Diagnose — workload is stuck, errored, or crash-looping
7	3. Observe — logs, traces, metrics, service stats for a running workload
8	4. Artifact lifecycle — iterate drafts, build images, lock for production, roll out new versions
9
10	## Prerequisites
11
12	`DATAROBOT_ENDPOINT` (must end in `/api/v2`) and `DATAROBOT_API_TOKEN` must be set. Run `datarobot-setup` if not. Auth header: `Authorization: Bearer ${DATAROBOT_API_TOKEN}`. The Workload API is not in the `datarobot` Python SDK — call REST directly.
13
14	Transport. Examples use Python `httpx` (`pip install httpx`). The API is plain HTTP, so equivalent calls work via `curl` or the `pulumi-datarobot` Pulumi provider declaratively. The skill teaches the model; transport is interchangeable.
15
16	## Bundled scripts
17
18	Runnable Python in `scripts/` (this skill's folder). Each uses `httpx` and reads `DATAROBOT_ENDPOINT` + `DATAROBOT_API_TOKEN`:
19
20	- `wait_for_running.py <workload_id>` — poll until `running`; exit 2 on terminal failure, 3 on timeout
21	- `diagnose_workload.py <workload_id>` — run the 5-step debug flow, print a structured diagnosis (`--json` for machine-readable)
22	- `wait_for_build.py <artifact_id> <build_id>` — poll a server-side image build; dumps last 2KB of logs on `FAILED`
23	- `wait_for_replacement.py <workload_id>` — poll a rolling replacement; handles the 404-when-cleared case
24	- `check_limits.py` — print the user's effective org-set scaling limits via `/account/info/`
25
26	## Deeper docs in references/
27
28	The SKILL.md is the operational core. Detail an agent needs occasionally lives in `references/`:
29
30	- `references/status-vocabulary.md` — workload + proton status enums and lifecycle transitions
31	- `references/common-error-patterns.md` — `CrashLoopBackOff` / `ImagePullBackOff` / `OOMKilled` / probe failures / `exec format error` / pending pods
32	- `references/schema-reference.md` — OpenAPI schemas worth looking up, credential type → key mappings, public-spec path-key quirks
33	- `references/lifecycle-flows.md` — artifact draft → lock → production flow rules and behavioral gotchas
34	- `references/code-to-workload.md` — deploy from source code (no Dockerfile authoring): `dr` CLI commands, `codeRef`, Execution Environments, generated vs provided Dockerfile modes, iterate-rebuild loop
35
36	## OpenAPI spec is source of truth
37
38	Published at `https://docs.datarobot.com/en/docs/api/reference/public-api/openapi.yaml`. The spec is ~5 MB — never dump it whole into context. Save once, then extract targeted slices with `yq`:
39
40	```bash
41	curl -sS "${DATAROBOT_ENDPOINT}/openapi.yaml" -o /tmp/wapi-spec.yaml
42	yq '.components.schemas.CreateWorkloadRequest' /tmp/wapi-spec.yaml
43	yq '.paths."/api/v2/workloads/{workload_id}/".patch' /tmp/wapi-spec.yaml
44	yq '.components.schemas \| keys \| .[]' /tmp/wapi-spec.yaml \| grep -i workload # discover
45	```
46
47	Python fallback: only `print()` the specific key, never the parsed dict. All workload paths in the public spec are keyed with `/api/v2/` prefix — see `references/schema-reference.md`.
48
49	---
50
51	# 1. Create / configure / scale
52
53	## Run a container as a workload (the 90% case)
54
55	```yaml
56	# spec.yaml — JSON also accepted; spec is sent verbatim
57	name: my-api-service
58	importance: low
59	artifact:
60	name: my-api-service-artifact
61	spec:
62	type: service
63	containerGroups:
64	- name: default
65	containers:
66	- name: main
67	imageUri: ghcr.io/org/my-app:latest
68	port: 8000
69	primary: true
70	readinessProbe: {path: /readyz, port: 8000, initialDelaySeconds: 10}
71	livenessProbe: {path: /healthz, port: 8000, initialDelaySeconds: 30}
72	runtime:
73	containerGroups:
74	- name: default # must match artifact.spec.containerGroups[].name (above)
75	replicaCount: 1
76	containers:
77	- name: main
78	resourceAllocation: {cpu: 1, memory: "512MB"}
79	```
80
81	```bash
82	dr workload create --spec-file spec.yaml # v0.2.74+; 4xx: 400=schema/li

People who liked this also viewed

6 picks

Type