Files
SentryAgent.ai Developer c93562e685 feat(phase-2): workstream 2 — Python SDK (sentryagent-idp)
Sync (requests) and async (httpx) clients with identical API surface
to the Node.js SDK.

Delivered:
- pyproject.toml — python>=3.9, hatchling build, mypy strict config
- types.py — all 14-endpoint request/response dataclasses
- errors.py — AgentIdPError with from_api_error, from_oauth2_error, network_error
- token_manager.py — thread-safe sync TokenManager, 60s refresh buffer
- async_token_manager.py — asyncio-safe AsyncTokenManager (httpx)
- _request.py — shared sync/async request helper (DRY)
- services/agents.py — AgentRegistryClient + AsyncAgentRegistryClient (5 methods each)
- services/credentials.py — CredentialClient + AsyncCredentialClient (4 methods each)
- services/token.py — TokenClient + AsyncTokenClient (introspect + revoke)
- services/audit.py — AuditClient + AsyncAuditClient (query + get)
- client.py — AgentIdPClient + AsyncAgentIdPClient
- __init__.py — barrel exports
- README.md — installation, quick start, full API reference

QA gates:
- mypy --strict: 0 errors (12 source files)
- pytest: 57/57 passed
- Coverage: 90.83% (required >= 80%)
- All 14 endpoints covered (sync + async)
- AgentIdPError raised on all failure paths

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-28 15:11:27 +00:00

215 lines
4.9 KiB
Markdown

# sentryagent-idp
Python SDK for the [SentryAgent.ai AgentIdP](https://sentryagent.ai) — the open-source Identity Provider for AI agents.
Handles token acquisition and caching automatically. Covers all 14 AgentIdP API endpoints. Provides both synchronous (`requests`) and asynchronous (`httpx`) clients.
---
## Requirements
- Python 3.9 or later
- A running AgentIdP server
- A registered agent with a valid `client_id` and `client_secret`
---
## Installation
```bash
pip install sentryagent-idp
```
---
## Quick start
### Synchronous
```python
from sentryagent_idp import AgentIdPClient
client = AgentIdPClient(
base_url="http://localhost:3000",
client_id="your-agent-id", # the agent's agentId (UUID)
client_secret="your-client-secret",
)
# List agents — token is acquired and cached automatically
result = client.agents.list_agents()
print(result.data)
```
### Asynchronous
```python
import asyncio
from sentryagent_idp import AsyncAgentIdPClient
async def main() -> None:
client = AsyncAgentIdPClient(
base_url="http://localhost:3000",
client_id="your-agent-id",
client_secret="your-client-secret",
)
result = await client.agents.list_agents()
print(result.data)
asyncio.run(main())
```
---
## Configuration
```python
client = AgentIdPClient(
base_url="http://localhost:3000",
client_id="a1b2c3d4-e5f6-7890-abcd-ef1234567890",
client_secret="your-client-secret",
# Optional: restrict scopes. Defaults to all four.
scopes=["agents:read", "tokens:read"],
)
```
| Parameter | Required | Description |
|-----------|----------|-------------|
| `base_url` | Yes | Base URL of the AgentIdP server |
| `client_id` | Yes | The agent's `agentId` (UUID) |
| `client_secret` | Yes | The credential secret |
| `scopes` | No | OAuth 2.0 scopes to request. Defaults to all four. |
---
## Token management
The SDK fetches and caches access tokens automatically. A new token is requested when the cached token is within 60 seconds of expiry.
```python
# Force a fresh token on the next request (e.g. after rotating credentials)
client.clear_token_cache()
```
---
## Agent Registry
```python
from sentryagent_idp import RegisterAgentRequest, UpdateAgentRequest
# Register a new agent
agent = client.agents.register_agent(RegisterAgentRequest(
email="classifier-v2@myorg.ai",
agent_type="classifier",
version="2.0.0",
capabilities=["text-classification", "sentiment-analysis"],
owner="platform-team",
deployment_env="production",
))
print(agent.agent_id) # UUID assigned by AgentIdP
# List agents
result = client.agents.list_agents(status="active", limit=20)
# Get a single agent
agent = client.agents.get_agent("a1b2c3d4-...")
# Update an agent
updated = client.agents.update_agent("a1b2c3d4-...", UpdateAgentRequest(
version="2.1.0",
capabilities=["text-classification", "sentiment-analysis", "intent-detection"],
))
# Decommission (irreversible)
client.agents.decommission_agent("a1b2c3d4-...")
```
---
## Credentials
```python
# Generate a credential — client_secret shown once, store it securely
cred = client.credentials.generate_credential("a1b2c3d4-...")
print(cred.client_secret) # only available here
# List credentials
result = client.credentials.list_credentials("a1b2c3d4-...")
# Rotate — same credential_id, new secret, old secret immediately invalid
rotated = client.credentials.rotate_credential("a1b2c3d4-...", "cred-uuid")
print(rotated.client_secret) # new secret — store immediately
# Revoke
client.credentials.revoke_credential("a1b2c3d4-...", "cred-uuid")
```
---
## Token operations
```python
# Introspect — check whether a token is active
result = client.tokens.introspect_token(some_token)
if result.active:
print(f"Token valid, expires at {result.exp}")
else:
print("Token is expired or revoked")
# Revoke — immediately invalidates the token
client.tokens.revoke_token(some_token)
```
---
## Audit log
```python
# Query audit events
result = client.audit.query_audit_log(
agent_id="a1b2c3d4-...",
action="token.issued",
outcome="success",
from_date="2026-03-01T00:00:00Z",
to_date="2026-03-31T23:59:59Z",
limit=50,
)
# Get a single event
event = client.audit.get_audit_event("event-uuid")
```
---
## Error handling
All API errors are raised as `AgentIdPError`:
```python
from sentryagent_idp import AgentIdPClient, AgentIdPError
try:
client.agents.get_agent("non-existent-id")
except AgentIdPError as err:
print(err.code) # e.g. "AgentNotFoundError"
print(err.http_status) # e.g. 404
print(str(err)) # human-readable description
```
---
## Available scopes
| Scope | What it allows |
|-------|----------------|
| `agents:read` | Read agent records |
| `agents:write` | Create, update, decommission agents |
| `tokens:read` | Introspect tokens |
| `audit:read` | Query audit logs |
---
## License
Apache 2.0 — see `LICENSE` in the repository root.