docs: fix rolling_trust_score description (it's not row-weighted)

[rachithga]

The streaming docs say rolling_trust_score is "row-weighted trust over the recent window," but that's not actually what the code does — it's just a plain mean of the last rolling_trust_window batch scores. So a 1-row batch ends up counting exactly as much as a 999-row batch, which isn't what "row-weighted" implies.

I checked git blame and the doc line and the implementation both came from the same original commit. So this isn't a regression, the docs have just been wrong since day one. The config docstring for rolling_trust_window already says "averaged," which matches the code, so this really is just a doc bug.

To confirm, I ran the repro from the issue:

• 1-row batch (all missing) → 0 trust
• 999-row batch (clean) → 100 trust
• rolling comes out to 50.0, not ~100 like row-weighting would give

cumulative_trust_score on the other hand IS genuinely row-weighted, so I left that line alone and only fixed the rolling_trust_score row.

Changes:

• Corrected the docs table row to describe what the metric actually does
• Added two tests pinning down the exact numbers from the issue's repro, so this doesn't quietly drift again
• Changelog entry

No changes to the actual trust-score logic. That'd be a real behavior change and probably deserves its own discussion if it's ever wanted.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 96bfdd18-5c53-44d7-9f7d-16b732932256

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

🔵 Needs a closer look

The key regression test does not currently exercise the stated 1-row scenario.

Pull request overview

Corrects the streaming documentation for rolling_trust_score and adds regression coverage without changing trust-score logic.

Changes:

• Updates the rolling score description.
• Adds weighting behavior tests.
• Records the fix in the changelog.

File summaries

File	Summary
tests/test_streaming_state.py	Adds weighting tests; the 1-row fixture currently drops the empty row and needs adjustment.
docs/streaming.md	Corrects the rolling score description.
CHANGELOG.md	Documents the fix.

Review details

Suppressed comments (1)

tests/test_streaming_state.py:114

• This case does not actually reproduce the stated 1-row batch: StreamingCleaner defaults to drop_empty_rows=True, and _build_rep_config leaves that setting enabled, so the all-None row is removed before rows_in_batch and record_trust are computed. As a result rows[0] is 0 rather than 1, and the test would continue to pass without exercising the issue's 1-vs-999 row weighting; disable empty-row dropping for this fixture (or use an equivalent non-dropped 1-row input).

    cleaner = StreamingCleaner(verbose=False)

• Files reviewed: 3/3 changed files
• Comments generated: 0
• Review effort level: Lite

---

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