awesome-repositories.com
Blog
MCP
awesome-repositories.com

Discover the best open-source repositories with AI-powered search.

ExploreCurated searchesOpen-source alternativesSelf-hosted softwareBlogSitemap
ProjectMCP serverAboutHow we rankPress
LegalPrivacyTerms
© 2026 Bringes Technology SRL·VAT RO45896025·hello@awesome-repositories.com
tigerless-labs avatar

tigerless-labs/agent-memory

0
View on GitHub↗

Agent Memory

Agent Memory gives AI coding agents durable long-term memory that persists across sessions. Each memory is kept as a plain Markdown file with a metadata header in a single shared location, so the store stays human-readable, portable, and searchable with ordinary tools. Several agents, such as Claude Code and Codex, can read and write the same store through lifecycle hooks, a skill, a shell command line, or an optional tool server. No API key or account is required.

Memory is captured automatically rather than left to the model's discretion. Session transcripts are distilled into structured memories at session start, stop, end, and compaction boundaries, using the host agent's own reasoning, and the full trace is copied first so any memory can be checked against the original messages. A separate background layer consolidates the store while idle: it merges, supersedes, deduplicates, and forgets entries. Risky unattended changes go through reviewable proposals with per-kind caps.

Corrections keep history. Memories can be revised, replaced, merged, linked, or retired, and old versions stay on disk with validity intervals, so a recall can answer what was true as of an earlier date. Edits run under a store-wide writer lock, and validation of names, identifiers, and configuration prevents writes outside the memory tree.

Retrieval is a local ranked search that returns one-line abstracts with paths and scores. Each entry can then be opened at the depth needed, from headings or an abstract to the full file or the cited raw messages. Optional embedding-based vector search is fused with keyword results, adding some latency on each search. Chinese text is segmented so Chinese queries match Chinese memories. The search index is a disposable cache that can be rebuilt from the Markdown files at any time.

Features

  • Markdown Memory Stores - Keeps every memory as a plain Markdown file with a metadata header, so the store is readable, portable, and searchable with ordinary tools.
  • Multi-Agent Memory Stores - Lets several coding agents such as Claude Code and Codex read and write one common memory store, so knowledge carries between tools.
  • Agent Memory Persistence - Gives coding agents long-term knowledge that survives across sessions, saved as plain files that later conversations can reuse.
  • Session Memory Recovery - Turns session transcripts into durable memories at session boundaries using the host agent's own reasoning after copying the full trace first.

AI search

Explore more awesome repositories

Describe what you need in plain English — the AI ranks thousands of curated open-source projects by relevance.

Start searching with AI
4,313 stars·205 forks·Python·MIT·1 view
  • Knowledge Distillation into Skills - Distils session transcripts into structured memories at session start, stop, end, and compaction, with links back to the original messages.
  • Memory Consolidation Workflows - Consolidates the store while idle by merging, superseding, deduplicating, and forgetting entries, with history kept and edits made under a writer lock.
  • Memory Supersession and Merging - Retires outdated memories by pointing to an active successor or merging named memories atomically, keeping old files and recording a replacement link.
  • Ranked Memory Retrieval - Runs local ranked search over stored memories with optional semantic vectors, Chinese text support, and opening entries at the depth needed.
  • Session Log Distillers - Distils session transcripts into structured memories automatically at session start, stop, end, and compaction events without relying on the agent to save.
  • Memory Search Engines - Runs a local ranked search over stored memories with adjustable limits and structured output, so agents can inspect hits and read entries.
  • Agent Memory Versioning - Keeps prior versions of revised, merged, or superseded memories with validity intervals so recall can answer what was true on a past date.
  • Automated Session Review - Routes unattended merge, split, supersede, and delete decisions through reviewable proposals with per-kind caps and validation before changes apply.
  • Depth-Tiered Reads - Opens a retrieved memory as headings, an abstract, or the full file so the agent pays only for the detail it needs.
  • Idle Memory Consolidation - Runs a separate background layer that merges, supersedes, deduplicates, and tidies stored memory during idle periods using authority tiers.
  • Memory Provenance Systems - Lets any memory be checked against the original session messages it was distilled from, showing source, roles, and times.
  • Memory Structuring - Writes typed memory entries with fields, an abstract, a body, and provenance, grouped into subdirectories defined by store schemas.
  • Conversational Memory Consolidation - Consolidates the memory store in the background while idle, merging, superseding, deduplicating, and forgetting entries, with risky changes sent to reviewable proposals.
  • Task Boundary Memory Writes - Save new memory deterministically when a defined task boundary is reached, instead of relying on the model to decide ad hoc when something is worth remembering.
  • Temporal Memory Queries - Keeps retired memories on disk with validity intervals so past facts stay searchable while current indexes list only still-valid memories.
  • Traceable Memory Correction - Corrects, replaces, merges, or deletes memories while keeping a chain of replaced versions for as-of-date recall.
  • Conversation Memory Retrieval - Searches the memory store for entries related to a query, opens the most relevant ones, and returns findings in one call before work starts.
  • Depth-Tiered Reads - Returns ranked abstracts with paths and scores, then opens each memory at the depth needed, down to the cited raw messages.
  • Index Rebuilds from Source Files - Treats the search index as a disposable cache that can be deleted and regenerated from the Markdown files at any time without losing memory.
  • AI-Based Relevance Ranking - Ranks memories locally by relevance and fuses embedding-based vector results with keyword matches, with Chinese text segmentation.
  • Memory Change Auditing - Answers what was true at an earlier date by keeping retired memories with validity intervals and tracing each memory back to raw evidence.

    • Star history

    Star history chart for tigerless-labs/agent-memoryStar history chart for tigerless-labs/agent-memory

    How this analysis was created: This summary and feature list are AI-generated from collected project material and can contain mistakes. Stars, license and language are imported from GitHub. Inclusion does not mean that we have tested or audited this project. Check the source documentation for any feature you depend on. Learn more on our About page.

    Frequently asked questions

    What does tigerless-labs/agent-memory do?

    Agent Memory gives AI coding agents durable long-term memory that persists across sessions. Each memory is kept as a plain Markdown file with a metadata header in a single shared location, so the store stays human-readable, portable, and searchable with ordinary tools. Several agents, such as Claude Code and Codex, can read and write the same store through lifecycle hooks, a skill, a shell command line, or an optional tool server. No API key or account is required.

    What are the main features of tigerless-labs/agent-memory?

    The main features of tigerless-labs/agent-memory are: Markdown Memory Stores, Multi-Agent Memory Stores, Agent Memory Persistence, Session Memory Recovery, Knowledge Distillation into Skills, Memory Consolidation Workflows, Memory Supersession and Merging, Ranked Memory Retrieval.

    Which projects share features with tigerless-labs/agent-memory?

    Projects with overlapping indexed features include: akitaonrails/ai-memory — ai-memory is a self-hosted long-term memory server for coding agents. It captures prompts, tool calls, and session… basedhardware/omi — Omi is an open-source wearable AI platform that captures audio and screen data to provide real-time conversational… rlaope/oh-my-hermes — Oh My Hermes (Hermes Orchestrator) is a workflow layer that sits on top of the Hermes agent runtime. It routes… cortexreach/memory-lancedb-pro — memory-lancedb-pro is a plugin that gives AI agents persistent long-term memory stored in LanceDB. It saves and… opensquilla/opensquilla — OpenSquilla is an LLM agent orchestration framework designed to coordinate multi-step AI workflows and tool execution… tencentcloud/tencentdb-agent-memory — TencentDB-Agent-Memory is a shared memory platform for large language model agents. It stores conversation facts,…

    Projects sharing features with Agent Memory

    These projects share indexed features with Agent Memory. Shared tags can include platform or build tooling; verify the primary use case before treating a result as a replacement.
    • akitaonrails/ai-memoryakitaonrails avatar

      akitaonrails/ai-memory

      15,330View on GitHub↗

      ai-memory is a self-hosted long-term memory server for coding agents. It captures prompts, tool calls, and session boundaries through lifecycle hooks and compiles them into searchable, git-versioned markdown wiki pages that later sessions can recall. It also acts as a Model Context Protocol server, exposing search, recall, and page read and write operations over standard input and HTTP streaming, and it can run without any language model configured, using rule-based summaries. What sets it apart is that it works across vendors. An installer wires hooks, tool-server entries, routing instructio

      Rust
      View on GitHub↗15,330
    • basedhardware/omiBasedHardware avatar

      BasedHardware/omi

      12,869View on GitHub↗

      Omi is an open-source wearable AI platform that captures audio and screen data to provide real-time conversational assistance and memory. It integrates a wearable hardware development kit with a vector memory database and large language model capabilities to create a persistent digital record of user interactions. The platform is distinguished by its BLE audio streaming pipeline, which transmits raw audio from wearable hardware for real-time transcription and speaker identification. It utilizes a plugin-based agent tool framework that allows AI assistants to autonomously invoke custom functio

      Dartaiappbci
      View on GitHub↗12,869
    • rlaope/oh-my-hermesrlaope avatar

      rlaope/oh-my-hermes

      5,229View on GitHub↗

      Oh My Hermes (Hermes Orchestrator) is a workflow layer that sits on top of the Hermes agent runtime. It routes requests to staged skills, external coding executors, and verification gates, through chat or a terminal interface. Completion is evidence-gated: every status carries a stage and confidence, and work is marked verified only after tests, reviews, or CI checks have actually passed. It is also distributed as a catalog of installable workflow skills and plugin hooks, for Hermes and other coding-agent hosts. Its main focus is multi-agent coding delegation. Scoped tasks are handed to codin

      Pythonagentaiai-agent
      View on GitHub↗5,229
    • cortexreach/memory-lancedb-proCortexReach avatar

      CortexReach/memory-lancedb-pro

      4,449View on GitHub↗

      memory-lancedb-pro is a plugin that gives AI agents persistent long-term memory stored in LanceDB. It saves and recalls facts, preferences, decisions, and entities across conversations, capturing new memories after each turn and injecting relevant ones into the agent's context before each reply, with no manual save command needed. What sets it apart is how memories are extracted and retrieved. A language model classifies conversation content into profile, preference, entity, event, case, and pattern categories, then deduplicates and merges results before storage. Recall blends vector similari

      JavaScriptlancedbmemoryopenclaw
      View on GitHub↗4,449
    Compare all 30 related projects→