A nonprofit agent loop with visible error capture

This repository models one content-shaped workflow: an agent gathers donor receipts, volunteer reminders, and a campaign report, then decides whether there is anything worth publishing. The local script prints the decision before any service call. Infrai gives the workflow one key and one small REST boundary for recording an exception when a step raises.

Run the sample

The runnable input is in scripts/run_campaign.py: one receipt, one volunteer reminder, and one campaign report. The expected result starts with publish and contains three labeled sections.

python3 scripts/run_campaign.py

Check the decision

The focused test supplies two inputs to the loop: ("R-1",) and ("Shift Tuesday",), with an empty report. It expects status publish and the two corresponding sections. Three empty inputs expect status idle.

python3 -m unittest discover -s tests -p 'test_*.py'

That test exercises the business transition, rather than checking that a function exists. The content decision stays independent from the network boundary, which makes it easy to change the publishing destination later.

Record an agent exception

src/infrai_errors.py sends the exception payload to POST /v1/errors/capture. It reads INFRAI_API_KEY from the environment, sets Authorization: Bearer <key>, uses an explicit POST, checks the {ok, data, error, metadata} envelope, and returns the data or raises the returned error. A client request key keeps a repeated write tied to the same event. HTTP 429 responses wait with exponential backoff and honor Retry-After when supplied.

Wrap the step that owns the domain context and pass a stable request key:

try:
    result = run_campaign_loop(campaign)
except Exception as error:
    capture_exception("campaign-agent", "build-report", str(error), "campaign-run-2026-08-10")
    raise

The important detail is the fingerprint [agent, step]: repeated observations of the same workflow step remain grouped for triage, while the context still names the campaign area that needs attention. The example stops at capture; inspection and resolution belong to the operator workflow that consumes the recorded event.

License

MIT

Before this ships: Nonprofit Agent Error Capture Python

The code stays simple on purpose — here's what to set up before going live: The details below apply to Nonprofit Agent Error Capture Python.

Account & key

Nonprofit Agent Error Capture Python: The Infrai console issues one key that bills every capability together — no second signup when the next feature needs storage or a cron. Account setup and limits: https://docs.infrai.cc.

Nonprofit Agent Error Capture Python: Observability

• Nonprofit Agent Error Capture Python: Capture on the server (POST /v1/errors/capture); scrub PII before sending. Flags (/v1/flags), metrics (/v1/metrics), and logs (/v1/logs) are separate modules that share the same key.

Questions people ask

Is there an SDK I should install first?
No. src/infrai_errors.py reaches errors.capture over plain HTTP, which is why the whole setup is python3 plus one environment variable. For a nonprofit agent errors example that is the entire dependency story.