ClawTrust LogoClawTrust
Vestige

Vestige

by Belkouche · v1.0.0

Marketing
ClawHub
8.1
/ 10
1 evaluations
2.6k Downloads

Overview

Vestige is a local, FSRS-6–based cognitive memory system that provides long-lived, decaying, and searchable memory across sessions using semantic + keyword search and spaced repetition-driven resurfacing.

Key Advantages

1.Persistent cross-session memory with natural decay that mimics human forgetting instead of storing everything forever.
2.Rich toolset (search, smart_ingest, ingest, codebase, intention, promote/demote_memory) for structured knowledge, preferences, and reminders.
3.Unified search combining keyword, semantic, and hybrid methods for more robust recall than plain text search.
4.Runs fully locally with clear data locations, improving privacy and making it suitable for sensitive or proprietary data.
5.Trigger-word integration (e.g., “remember this”, “remind me…”) that can be wired into an agent’s prompt logic for more natural memory usage.

Use Cases

  • Long-term user preference storage (technology stack choices, UI/theme preferences, communication style).
  • Persisting project and codebase context across sessions, including architectural decisions and recurring patterns.
  • Remembering bug fixes, workarounds, and troubleshooting steps that should surface when similar issues reoccur.
  • Task and reminder-style use via the intention tool for future triggers and follow-ups.
  • Enhanced semantic recall for important facts or notes that are hard to retrieve via simple keyword search.

Evaluation Scores

8.1
/ 10
Reliability
7.5
Functionality
9.0
Usability
7.5
Safety
8.5
Performance
8.0
Compatibility
7.8

Based on 1 evaluation · Latest: 3/19/2026

Download Trend

Loading...

Evaluation History (1)

8.1/103/19/2026
▼
OS: darwin-x64LLM: google/gemini-2.5-flash
**Judgement:** Vestige is a strong, developer-oriented local memory layer that gives an agent persistent, semantically searchable memory with realistic decay and spaced repetition. It is especially well-suited for technical users willing to install and manage local binaries. **Recommended scenarios** - You want robust, cross-session memory for projects, preferences, and solutions, beyond simple Markdown logs. - You work with sensitive or proprietary data and prefer fully local storage and embeddings. - You build agents that should “remember” user preferences and important facts while still allowing unused information to fade. **Key strengths** - Rich, FSRS-6–based memory model with semantic + keyword search and tools for ingest, promotion, demotion, and reminders. - Fully local operation with explicit data and cache paths, good for privacy-conscious setups. - Clear trigger phrases and integration patterns that agents can hook into. **Risks / limitations** - Requires local binaries and file-system access; setup and maintenance may be non-trivial and less suitable for non-technical users or constrained environments. - Reliability characteristics (backups, corruption handling, large-scale performance) are not clearly documented and should be validated before relying on it for critical knowledge. - Windows support is not mentioned; cross-platform behavior may be uneven, reducing out-of-the-box compatibility.

Comments (0)

Post a Comment

No comments yet. Be the first!