feat(connectors-sdk): expose live stream recovery params in BaseStreamConnectorConfig (#7590)

[Thibaut Rouxel (throuxel)]

Proposed changes

• Expose the stream "recovery mode" in BaseStreamConnectorConfig so it is generated in every SDK-based stream connector's config schema (previously only settable via raw env vars, invisible in the UI / XTM Composer). New fields:
  • live_stream_start_timestamp (int) — stream position to start from, as epoch milliseconds; a field_validator enforces 13 digits so a seconds value raises a ValidationError instead of silently replaying the whole stream from the beginning.
  • live_stream_recover (bool, default true) — whether to replay historical events (recover/backfill) on first start. Enabled by default (current behaviour: full backfill up to now). Set to false to only process new events.
  • live_stream_recover_iso_date (date-time) — optional cutoff date up to which history is replayed when recover is enabled.
• Keep the generated schema UI-friendly: three single-typed properties (integer / boolean / string+date-time), no anyOf.
• Bridge to pycti via a scoped serializer: when serializing for the helper, live_stream_recover=false maps live_stream_recover_iso_date to the none keyword pycti expects (disables ?recover=); otherwise the date (or nothing) is passed through. Non-pycti dumps keep the raw values (fidelity / round-trip safe).
• Add unit tests (defaults, seconds-timestamp rejection, recover-off -> none, date serialization, schema exposure).

Related issues

• Fixes feat(connectors-sdk): expose live stream recovery params in BaseStreamConnectorConfig #7590

Checklist

• I consider the submitted work as finished
• I have signed my commits using GPG key.
• I tested the code for its functionality using different use cases
• I added/update the relevant documentation (either on github or on notion)
• Where necessary I refactored code to improve the overall quality

Further comments

These variables are read directly by pycti (opencti_connector_helper.py) but were not declared in the SDK, so they were absent from the auto-generated connector_config_schema.json. They are only applied on a connector's first run (empty state); afterwards start_from / recover_until are persisted in the connector state.

The config schemas of SDK-based stream connectors are intentionally not regenerated here — the CI regenerates them.

[Copilot]

Pull request overview

This PR exposes live-stream recovery settings in the SDK base configuration with validation and pycti-specific serialization.

Changes:

• Adds recovery timestamp and ISO date fields.
• Validates millisecond timestamps.
• Adds serialization, default, rejection, and schema tests.

Review findings:

• Critical (3 votes): Preserve existing no and none recovery sentinels.
• Nit (2 votes): Add coverage for True serialization returning None.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Summary
connectors-sdk/tests/test_settings/test_base_settings.py	Tests recovery defaults, validation, serialization, and schema exposure.
connectors-sdk/connectors_sdk/settings/base_settings.py	Defines and validates recovery configuration fields and serialization.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Helene Nguyen (helene-nguyen)]

Thibaut Rouxel (@throuxel) All good to me, thanks !