Ship: verified-memory gate on by default; preset ladder (Quality gate on, Integrity strict)

[jaylfc]

Finalizes the two ship decisions you signed off on.

The flip was already live at the bare default. config.get_controls falls back to prefer_verified (on) when nothing is persisted, and get_runtime_overrides excludes unset controls so the recommended recipe's reranker stays on. So this PR makes the presets honour 'on across tiers' and records the decision; it does not change the out-of-the-box default (which was already on).

Decision 1 (prefer_verified flip on). Backed by E-018: served-hallucination 0.040 to 0.000 under all three judge families (llama / qwen-instruct / gemma) at n=250, at no measured accuracy or recall cost. The preset ladder is now three clear rungs:

Preset	gate	meaning
Minimal	off	lean opt-out, no rerank
Quality	prefer_verified	recommended accuracy tier, gate on
Integrity	strict	maximum purity: drops unverified claims, accepts a small measured recall trade-off

This resolves the prior inconsistency where Quality (the common accuracy pick) silently turned the gate off. profiles.resolve_config now accepts an explicit pinned value in a profile override (so Integrity can write strict), in addition to the existing bool switches.

Decision 2 (E-012 ship: rerank default + self-verify opt-in). No code change was needed: reranking is already on in the recommended retrieve recipe and in the Quality/Integrity presets, and answer self-verification stays an opt-in consumer-side recommendation enabled in the Quality and Integrity install profiles (taOSmd serves memory; answer generation is the consumer's path). Documented as such.

Updated controls.py, profiles.py, the README and INTEGRATION preset tables, the CHANGELOG, and the profile/controls/http tests. Full suite 970 pass. The research-report status flip (F-011/E-018 pending-Jay to signed-off) will follow on merge.

Summary by CodeRabbit

Release Notes

• Documentation

  • Updated changelog, README, and integration documentation to reflect default behavior of the verification recall gate
  • Quality tier preset now enables the verification gate by default (previously disabled)
  • Integrity tier preset uses a stricter verification variant
  • Minimal tier preset disables the verification gate
  • Cross-encoder reranker remains enabled in the recommended configuration

• Tests

  • Updated tests to validate new preset configurations and gate behavior across tiers

[qodo-code-review]

Qodo reviews are paused for this user.

Troubleshooting steps vary by plan Learn more →

On a Teams plan?
Reviews resume once this user has a paid seat and their Git account is linked in Qodo.
Link Git account →

Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center?
These require an Enterprise plan - Contact us
Contact us →

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 0a23b0e2-3cce-4738-84ff-06b7d69e6434

📥 Commits

Reviewing files that changed from the base of the PR and between d665eb7 and a8657f9.

📒 Files selected for processing (7)

• CHANGELOG.md
• README.md
• docs/INTEGRATION-memory-config.md
• taosmd/controls.py
• taosmd/profiles.py
• tests/test_http_server.py
• tests/test_profiles.py

---

📝 Walkthrough

Walkthrough

Enables prefer_verified as the global default recall gate. Updates PRESETS["quality"] to turn it on and PRESETS["integrity"] to use the "strict" variant. Extends resolve_config in profiles.py to handle string-valued profile overrides as pinned config values under existing consent rules. Tests and documentation are updated to match.

Changes

prefer_verified global default and preset ladder

Layer / File(s)	Summary
PRESETS definitions and resolve_config string-override logic taosmd/controls.py, taosmd/profiles.py	PRESETS["quality"] sets prefer_verified to "prefer_verified" (was "off"); PRESETS["integrity"] sets it to "strict" (was "prefer_verified"). The integrity profile override changes from boolean True to string "strict". resolve_config gains a new branch: when an override value is a string, it writes that pinned value directly if consent rules allow, otherwise writes off_value and skips the boolean enable path.
Tests for strict gate and quality recall gate tests/test_http_server.py, tests/test_profiles.py	Integrity preset roundtrip test asserts prefer_verified == "strict". The old integrity resolve_config test is replaced with test_resolve_config_integrity_uses_strict_gate (asserting "strict"), and test_resolve_config_quality_enables_recall_gate is added for the quality tier.
Documentation: preset ladder and global default CHANGELOG.md, README.md, docs/INTEGRATION-memory-config.md	CHANGELOG documents the global prefer_verified default and the Minimal/Quality/Integrity preset ladder. Preset tables in README and integration docs are updated to show Quality on and Integrity strict.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• jaylfc/taosmd#168: Modifies taosmd/profiles.py around resolve_config and profile override handling for switch values, including consent-gated behavior and pinned string overrides — directly related to the resolve_config string-pinning logic introduced here.

Poem

🐇 The gate swings open by default now, friend,
Quality keeps it on, Integrity goes strict to the end.
No more "off" for the memory that's true —
Pinned strings flow through consent like morning dew.
The hallucinations shrink, the recall stays bright,
A verified bunny hops into the light! ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 20.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main changes: enabling the verified-memory gate by default and restructuring the preset ladder with specific behaviors for Quality and Integrity presets.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch ship/preset-ladder

---

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 and usage tips.}

[gitar-bot]

Note

Your trial team has used its Gitar budget, so automatic reviews are paused. Upgrade now to unlock full capacity. Comment "Gitar review" to trigger a review manually.
Learn more about usage limits

Code Review ✅ Approved

Enables the verified-memory gate by default across Quality and Integrity presets, establishing a clear ladder for memory safety and accuracy tiers. All configuration overrides, documentation, and tests successfully updated to support the new preset hierarchy.

Options

Display: compact → Showing less information.

Comment with these commands to change:

Compact

gitar display:verbose

---

Important

Your trial ends in 4 days — upgrade now to keep code review, CI analysis, auto-apply, custom automations, and more.

_{Was this helpful? React with 👍 / 👎 | Gitar}