Appearance
Agents
Specialist agents with a standard input/output contract. Agents use the same /v1 capabilities, jobs, and artifacts as the rest of the platform.
GET /v1/agents
List registered specialists.
bash
curl -s https://api.skilleo.org/v1/agents \
-H "Authorization: Bearer $SKILLEO_API_KEY"Each agent includes id, name, description, status, capability, input_contract, and output_contract.
GET /v1/agents/{id}
Return one specialist (e.g. /v1/agents/image.generator).
bash
curl -s https://api.skilleo.org/v1/agents/image.generator \
-H "Authorization: Bearer $SKILLEO_API_KEY"| Code | When |
|---|---|
NOT_FOUND | Unknown agent id |
POST /v1/agents/{id}/runs
Start an agent run. Returns 202 with an agt_ id.
Request body
| Field | Type | Description |
|---|---|---|
objective | string | Required. What to produce |
inputs | array | Optional reference artifacts: {artifact_id, role} |
constraints | object | Specialist-specific fields (e.g. prompt, width, height, steps) |
budget | object | Optional max_cost_usd |
available_capabilities | array | Optional allow-list; must include the agent's capability when set |
compute_id | string | Optional execution target |
endpoint_id | string | Optional execution target |
Example
bash
curl -s -X POST https://api.skilleo.org/v1/agents/image.generator/runs \
-H "Authorization: Bearer $SKILLEO_API_KEY" \
-H "Content-Type: application/json" \
-d '{"objective":"A red cube product photo on white background","constraints":{"width":512,"height":512},"budget":{"max_cost_usd":1.0},"available_capabilities":["image.generate"]}'GET /v1/agent-runs/{id}
Poll run status. When completed, data.result holds the standard output contract:
bash
curl -s https://api.skilleo.org/v1/agent-runs/agt_abc123 \
-H "Authorization: Bearer $SKILLEO_API_KEY"| Field | Description |
|---|---|
status | completed or failed |
artifacts | Output artifacts with id, role, type, mime_type |
evidence | Trace objects (job id, prompt, input artifacts) |
metrics | inference_ms, total_ms |
cost | Estimated USD |
First specialist: image.generator
Text-to-image specialist using the image.generate capability. Returns artifacts in the standard output contract.
SDK
python
run = client.agents.run(
"image.generator",
objective="A red cube on white",
constraints={"width": 512},
)
done = client.agents.wait(run["id"])
artifact_id = done["data"]["result"]["artifacts"][0]["id"]
client.artifacts.save(artifact_id, "out.png")