StrataGate

Keep recent conversations verbatim. Show older history as an index. Answer only when the evidence is sufficient.

A layered memory and evidence retrieval system for long-running AI agents.

中文说明 · Architecture · Full evaluation

DeepSeek Harness plugin: automatic, local-first cross-session memory that remembers user preferences, project decisions, completed conversations, and tool results. It checks recalled evidence and can trace it back to the original messages before the agent answers. Install stratagate-dsh; implementation and usage details are in integrations/deepseek-harness.

LoCoMo conv-26: StrataGate averaged 80.46% accuracy across 10 independent Judge runs, versus 63.22% for Mem0 base (+17.24 percentage points)

Majority-correct: 121 / 152 vs 96 / 152 (+25 questions)

What problem does StrataGate solve?

A long-running agent needs more than a way to “store more.” When it answers, it must retrieve evidence that is correct, complete, and verifiable.

Keeping only summaries can lose dates, qualifications, and original wording. Similarity search can return related material that belongs to a different event. Treating every search hit as useful memory can also create a self-reinforcing retrieval loop.

StrataGate designs long-term memory around four core problems:

Common problem	How StrataGate handles it
History keeps growing and no longer fits in context	Store conversations as L0–L5 layered views; older memories default to shallower levels
A summary omits a date, exact wording, or qualification	Preserve the L5 source messages permanently, so every derived memory can return to its source
Search finds related material, but not enough evidence to answer	Use an evidence gate to judge sufficiency; if evidence is incomplete, change strategy, expand an event, or inspect the source
Frequently retrieved results keep reinforcing themselves	Update long-term weight only for memories that the final answer actually uses

StrataGate is not designed to make an agent retrieve more on every turn. It is designed to make the agent know whether the current evidence is sufficient and where to look next.

Experimental results

The current public comparison covers LoCoMo conv-26:

• 419 messages;
• 35 sessions;
• 152 category 1–4 questions;
• 10 independent Judge evaluations per question.

Metric	StrataGate	Mem0 base	Difference
Mean accuracy across 10 Judge runs	80.46%	63.22%	+17.24 percentage points
Majority-correct	121 / 152 (79.61%)	96 / 152 (63.16%)	+25 questions
Temporal	74.86%	34.59%	+40.27 percentage points
Single-hop	89.29%	75.14%	+14.14 percentage points
Multi-hop	66.56%	61.56%	+5.00 percentage points
Open-domain	83.08%	84.62%	-1.54 percentage points

The largest difference is in temporal questions. This is consistent with StrataGate's design—explicit event occurrence times, preserved source timestamps, and raw-source verification—but it is not a single-component ablation, so the full gap cannot be attributed to one field or retrieval step.

…