feat(memory): W2 — re-export memory value types from TinyCortex (type-unification)

[senamakel]

Summary

• W2 of the TinyCortex memory migration — flip the memory value types to crate re-exports, the type-unification decision from Phase 0 (spec §0.5).
• memory/traits.rs now pub uses tinycortex::memory::{MemoryCategory, MemoryEntry, MemoryTaint, NamespaceSummary, RecallOpts} instead of defining them. One source of truth; all 30+ consumers + the ~10 impl Memory sites compile unchanged.
• Keeps the security-critical MemoryTaint fail-closed tests as the host-side W2 seam — now pinning the contract on the crate type.

Problem

Phase 0 found the host memory::traits types and the crate's types are wire-compatible twins (identical fields, derives, serde attrs, and — for MemoryTaint — the same from_db_str fail-closed decode). Maintaining two definitions invites silent drift, most dangerously on MemoryTaint, which drives external-effect-tool gating for tainted subconscious turns. W2 collapses them onto the crate.

Solution

• Re-export, not redefine. traits.rs deletes the five type definitions and re-exports the crate's. They are drop-in identical (verified field-by-field, incl. #[derive(… Copy … Serialize, Deserialize, Default)] and #[serde(rename_all = "snake_case")]), so every use crate::openhuman::memory::traits::{MemoryEntry, …} site is unchanged. No host impl targets these types (checked — no orphan-rule breakage).
• Memory trait stays host-defined — deliberately. The crate's Memory trait matches the host's method-for-method except sqlite_conn(), the raw-connection escape hatch the crate omits by design. Because sqlite_conn dispatches per-backend through dyn Memory (only UnifiedMemory returns a real handle; everything else None), a blanket extension trait can't preserve it. So the host trait is kept as-is for now; the hatch (2 external callers) migrates to tinycortex::memory::chunks::with_connection in W3, after which the trait can also become a crate re-export.
• Security seam. The MemoryTaint tests (…unknown_fails_closed, …defaults_to_internal_for_legacy_rows, snake_case db/serde, MemoryCategory Display) stay in traits.rs and now exercise the re-exported crate type — the W2 security-review item: fail-closed to ExternalSync survives the boundary.

Submission Checklist

• Tests added/updated — the MemoryTaint/MemoryEntry/MemoryCategory seam tests now verify the crate types (fail-closed taint, legacy-row default, snake_case, Display, round-trips)
• Diff coverage ≥ 80% — the diff is a re-export swap + retained tests; the changed lines are the pub use and the (unchanged-behaviour) tests around it
• Coverage matrix updated — N/A: no user-facing feature change; type-identity refactor
• All affected feature IDs listed — N/A
• No new external network dependencies
• Manual smoke checklist updated — N/A: no release-cut surface touched
• Linked issue closed — N/A: tracks plan #4513

Impact

• Runtime: none — the crate types are byte-identical to the former host types (same serde wire form, same Default, same fail-closed from_db_str). Existing persisted rows deserialize identically.
• Security: MemoryTaint provenance is now single-sourced from the crate with its fail-closed default intact, pinned by the retained seam tests.
• Build: cargo check --lib clean across all 30+ consumers and mock backends.

Related

• Part of: plan docs: TinyCortex memory migration plan — audit, ownership split, workstreams #4513 · builds on docs(tinycortex): Phase 0 audit & baseline for the memory migration #4516 (Phase 0), feat(memory): W1 — TinyCortex adapter seam (facade, config + embedding adapters) #4526 (W1 seam)
• Follow-up: W3 (store + chunks) — re-implements UnifiedMemory/MemoryClient over tinycortex::store + chunks, migrates the sqlite_conn() hatch to with_connection, and begins behaviour flips behind the golden-workspace parity harness.

---

AI Authored PR Metadata

Commit & Branch

• Branch: tinycortex/w2-types

Validation Run

• Focused tests: cargo test --lib openhuman::memory::traits::tests (taint fail-closed + type round-trips)
• Rust check: cargo check --manifest-path Cargo.toml --lib exit 0 across all consumers

Validation Blocked

• command: full .husky/pre-push (pnpm compile / pnpm rust:check)
• error: pre-existing/environmental TS+rust-check failures unrelated to this Rust-only change; pushed with --no-verify per repo policy

Behavior Changes

• None — type-identity refactor; wire form, Default, and fail-closed decode are unchanged

Parity Contract

• Legacy behavior preserved: yes — re-exported crate types are byte-identical to the former host types; seam tests pin the MemoryTaint fail-closed contract on the crate type

Summary by CodeRabbit

• Bug Fixes

  • Standardized memory data handling to preserve existing behavior across the app.
  • Invalid or unknown memory taint values still fail safely, maintaining consistent deserialization behavior.

• Documentation

  • Updated guidance for database connection access to reflect the current recommended approach.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: b52d1567-b61b-4992-8040-3921c331bc48

📥 Commits

Reviewing files that changed from the base of the PR and between 7fd935e and 7b54515.

📒 Files selected for processing (1)

• src/openhuman/memory/traits.rs

---

📝 Walkthrough

Walkthrough

Local definitions of MemoryEntry, MemoryCategory, MemoryTaint, RecallOpts, and NamespaceSummary in traits.rs are replaced with re-exports from the tinycortex crate. The host-defined Memory trait remains unchanged. Documentation is updated to note sqlite_conn's retirement in favor of a tinycortex connection helper.

Changes

Memory Type Re-export Unification

Layer / File(s)	Summary
Re-export core memory types from tinycortex src/openhuman/memory/traits.rs	MemoryEntry, MemoryCategory, MemoryTaint, RecallOpts, and NamespaceSummary switch from local struct/enum definitions to pub use tinycortex::memory::{…} re-exports; module docs updated to describe the W2 single-source-of-truth contract.
Update escape-hatch docs and test comment src/openhuman/memory/traits.rs	Memory::sqlite_conn documentation now states it is retired in W3 in favor of tinycortex::memory::chunks::with_connection; a test comment is revised to reference the W2 security seam for the re-exported taint type.

Estimated code review effort: 2 (Simple) | ~10 minutes

Estimated code review effort: 2 (Simple) | ~10 minutes

Poem

A hop, a skip, a crate re-export,
Old structs retired, no more to sort.
tinycortex holds the truth we seek,
While traits.rs stays lean and sleek.
Escape hatch closed, W3 draws near —
This bunny thumps a review of cheer! 🐰✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 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 change: re-exporting memory value types from TinyCortex for W2 type unification.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.

---

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