Reviews Directory

Type: kb/types/index.md

← Parent

  • A-MEM (agent-memory-system-review) - A-MEM review: Zettelkasten-inspired Chroma memory notes with LLM evolution, fixed operations, pull retrieval, and weak metadata-generation wiring
  • ACE (agent-memory-system-review) - Playbook-learning loop that turns benchmark attempts and reflections into prompt-injected bullets with counters, append-only curation, and optional duplicate merging
  • Agent Skills for Context Engineering (agent-memory-system-review) - Static Agent Skills collection for context engineering: plugin-packaged instructional artifacts with progressive disclosure, activation triggers, examples, and no runtime memory learner
  • Agent Workflow Memory (agent-memory-system-review) - Workflow-memory system for web agents that distills WebArena and Mind2Web trajectories into website-scoped prompt files with weak lineage and lifecycle controls
  • Agent-R (agent-memory-system-review) - MCTS self-training pipeline that mines environment rollouts into verifier-spliced revision conversations, then documents external fine-tuning as the behavior-learning step
  • Agent-S (agent-memory-system-review) - Agent-S review: Simular GUI-agent stack with S1/S2 JSON experience memory, S2.5/S3 prompt-time reflection, code-agent delegation, and BBoN trace judging
  • AgentFly (agent-memory-system-review) - Memento planner-executor agent that learns from judged benchmark runs by appending plan cases to JSONL memory and optionally training a neural case selector
  • Agentic Harness Engineering (agent-memory-system-review) - Observability-driven coding-agent harness evolution loop that turns benchmark traces and debugger analyses into evaluated prompt, tool, middleware, skill, and memory edits
  • Amazon Science SAGE (agent-memory-system-review) - Amazon Science SAGE review: AppWorld traces, generated Python skills, skill-integrated rewards, SFT data, and GRPO checkpoints as distinct memory artifacts
  • Archie (agent-memory-system-review) - Arch Linux workstation configuration repo with Stow deployment, agent-readable work items, and session-log documentation but no automated memory-learning loop
  • AriGraph (agent-memory-system-review) - TextWorld graph-memory agent that converts observations into an in-run triplet graph plus episodic store for planning, retrieval, exploration, and navigation
  • ARIS (agent-memory-system-review) - ARIS review: markdown skill-pack for autonomous research with project state files, research wiki, review traces, and hook-log-driven skill optimization
  • Ars Contexta (agent-memory-system-review) - Claude Code plugin that derives agent-operated markdown knowledge systems with generated context, skills, hooks, queues, methodology notes, and friction-mining loops
  • Atomic (agent-memory-system-review) - Database-backed markdown atom KB with derived chunks, embeddings, tags, semantic edges, generated wiki articles, chat, and MCP tools
  • auto-harness (agent-memory-system-review) - Minimal benchmark-driven coding-agent improvement loop with trace reading, mutable agent code, regression-suite promotion, result lineage, and learnings logs
  • Autocontext (agent-memory-system-review) - Multi-role iterative control plane that turns scenario runs, runtime sessions, and production traces into playbooks, lessons, gates, harness mutations, datasets, and optional local model checkpoints
  • Awesome Agent Memory (agent-memory-system-review) - Review of Awesome Agent Memory as a curated README bibliography of memory products, benchmarks, papers, and articles rather than an implemented memory runtime
  • Binder (agent-memory-system-review) - Local-first SQLite knowledge graph whose markdown files are editable projections over typed entities, transactions, schemas, views, LSP, CLI, HTTP, and MCP
  • browzy.ai (agent-memory-system-review) - Terminal TypeScript personal KB that compiles raw sources into a markdown wiki, serves Q&A through SQLite FTS and section retrieval, and writes thin session-derived digests and insight drafts
  • ByteRover CLI (agent-memory-system-review) - ByteRover CLI review: file-backed context tree with curate/query agents, BM25 retrieval, summaries, manifests, VC, review gates, MCP, and trace-derived runtime ranking
  • cass_memory_system (agent-memory-system-review) - Coding-agent procedural memory CLI that mines cass session history into diary JSON, scored YAML playbook rules, outcomes, trauma guards, and MCP/CLI context surfaces
  • Claude Context Guard (agent-memory-system-review) - Claude Context Guard review: Claude Code safeguard files, slash-command recovery workflows, hooks, archives, and manual trace-to-ledger continuity
  • ClawVault (agent-memory-system-review) - Deprecated OpenClaw-era markdown memory system with session ledgers, scored observations, fact stores, context profiles, hooks, and maintenance workers
  • Closure-SDK (agent-memory-system-review) - Closure-SDK review: geometric integrity SDK plus Closure DNA and a runtime brain whose genome learns from ordered experience and feedback traces
  • Cludebot (agent-memory-system-review) - Cognitive memory SDK and Cludebot service with typed memories, hybrid retrieval, dream consolidation, action-outcome learning, graph links, MCP tools, and hosted/self-hosted/local modes
  • CocoIndex (agent-memory-system-review) - Incremental Python/Rust indexing engine whose flow definitions maintain derived vector, graph, table, and file targets from source artifacts
  • Cognee (agent-memory-system-review) - Cognee review: pipeline-first memory engine with add/cognify/memify/search flows, DataPoint schemas, poly-store backends, ACLs, and trace-to-graph improvement
  • Context Constitution (agent-memory-system-review) - Letta's prose constitution for memory-native agents, defining token-space learning policy, context ownership, MemFS affordances, and agent self-governance
  • CORAL (agent-memory-system-review) - CORAL review: multi-agent coding harness with shared attempts, notes, skills, heartbeat prompts, and score-driven trace-to-artifact learning
  • cq (agent-memory-system-review) - Mozilla.ai shared agent knowledge commons with SQLite local/private stores, MCP query/propose/confirm/flag tools, review gates, and prompt-mediated session reflection
  • CrewAI Memory (agent-memory-system-review) - CrewAI review: unified in-framework memory with scoped vector records, task-output extraction, prompt injection, agent memory tools, and HITL lesson learning
  • Decapod (agent-memory-system-review) - Repo-native coding-agent governance kernel with embedded constitution, .decapod state, proof gates, context capsules, typed memory, and governed trace-derived promotion
  • DocMason (agent-memory-system-review) - Repo-native document research app with file-only KB publication, canonical ask ownership, provenance traces, admissibility gates, and promoted interaction memories
  • Dynamic Cheatsheet (agent-memory-system-review) - Test-time learning benchmark that carries a curator-rewritten cheatsheet across ordered problem attempts and can retrieve prior examples by embedding similarity
  • engraph (agent-memory-system-review) - Obsidian vault server with canonical markdown files, SQLite FTS/vector/graph indexes, hybrid ranking, MCP/HTTP write tools, health diagnostics, and narrow placement-feedback learning
  • EQUIPA (agent-memory-system-review) - Multi-agent coding orchestrator with SQLite run memory, worktree isolation, trace-derived lessons/rules, prompt/config mutation loops, and optional training export
  • Exocomp (agent-memory-system-review) - Local Go coding-agent harness with role prompts, subprocess agents, tool allowlists, sandbox path checks, coordination JSON, and loadable skills
  • ExpeL (agent-memory-system-review) - ExpeL review: staged trajectory gathering, rule mutation, and prompt-time trajectory retrieval for benchmark agent learning
  • G-Memory (agent-memory-system-review) - G-Memory review: multi-agent trace capture, Chroma task recall, task-neighborhood retrieval, and scored insight rules with role-projected prompt use
  • GBrain (agent-memory-system-review) - Personal-brain CLI and MCP runtime with markdown-canonical pages, Postgres/PGLite indexes, trace-derived synthesis, skills, recipes, and maintenance cycles
  • getsentry/skills (agent-memory-system-review) - Sentry's Agent Skills marketplace: a file-backed skill library with provenance contracts, progressive disclosure, validators, and meta-skills for writing and scanning skills
  • Gnosis (agent-memory-system-review) - Review of Gnosis, a repo-local CLI memory system where agents search and write JSONL decision entries through doctrine-mediated workflow
  • Graphiti (agent-memory-system-review) - Graphiti review: temporal context graph with episode provenance, LLM extraction, bi-temporal fact edges, graph backends, and pull-only retrieval
  • HALO (agent-memory-system-review) - HALO review: trace-analysis engine over OTel JSONL with sidecar indexes, bounded trace tools, sandboxed analysis, subagents, and external harness-patch promotion
  • Hindsight (agent-memory-system-review) - Hindsight review: production memory backend with trace-to-fact retain, PostgreSQL/vector indexes, observations, directives, reflect tools, tenancy, and embedded mode
  • HippoRAG (agent-memory-system-review) - HippoRAG review: graph-augmented RAG library that turns corpus documents into OpenIE triples, embeddings, igraph state, and PPR retrieval activation
  • Hyalo (agent-memory-system-review) - Hyalo review: Rust CLI control plane for markdown/Obsidian vaults with scanning, indexes, link repair, saved views, hints, schemas, and Claude skills
  • HyperAgents (agent-memory-system-review) - HyperAgents review: self-editing agent harness where benchmark traces and scores select executable git-diff patches rather than prose memories
  • KBLaM (agent-memory-system-review) - KBLaM review: Microsoft Research model fork that projects key-value KB rows into learned attention key/value tensors rather than a retrievable agent KB
  • LACP (agent-memory-system-review) - Local agent control plane with routed execution, Claude hooks, Obsidian memory automation, receipts, provenance, and operational gates
  • Letta (agent-memory-system-review) - Letta review: agent-self-managed memory with core blocks, archival passages, recall search, compaction summaries, sleeptime editing, and git-backed block history
  • LLM Wiki (agent-memory-system-review) - LLM Wiki review: promptware knowledge-base protocol with topic-isolated markdown wikis, raw-to-compiled workflows, multi-runtime plugins, and local lint validation
  • LLM Wiki (kenhuangus) (agent-memory-system-review) - Executable LLM Wiki review: source monitors, local LLM extraction, markdown wiki storage, BM25 search, linting, UI/API, metrics, and incomplete prompt optimization
  • MehmetGoekce/llm-wiki (agent-memory-system-review) - Claude Code promptware scaffold for an L1/L2 Logseq or Obsidian wiki, with setup templates, slash-command workflows, schema, lint, and OpenSpec contracts
  • Mem0 (agent-memory-system-review) - Mem0 review: V3 ADD-only trace extraction into vector memories, entity-link retrieval boosts, plugin hooks, and legacy CRUD prompt remnants
  • Memori (agent-memory-system-review) - Memori review: SDK and cloud/BYODB memory layer that captures LLM turns, distills facts/triples/summaries, and injects recalled context
  • MemoryOS (agent-memory-system-review) - MemoryOS review: hierarchical conversational memory that stores dialogue traces, condenses sessions, extracts profile and knowledge records, and injects retrieved context
  • MemPalace (agent-memory-system-review) - MemPalace review: local Chroma drawer memory with conversation mining, hooks, MCP tools, diaries, temporal SQLite KG, and retrieval-first activation
  • MentisDB (agent-memory-system-review) - MentisDB review: Rust append-only thought ledger with hash-chain integrity, storage adapters, ranked retrieval, MCP/REST/dashboard surfaces, and signed skill registry
  • Meta-Harness (agent-memory-system-review) - Stanford IRIS Lab framework for trace-derived harness optimization with onboarding specs, Claude-proposed candidate files, benchmark gates, and frontier artifacts
  • MiroShark (agent-memory-system-review) - Document-to-social-simulation stack with Neo4j graph extraction, cross-platform round memory, belief drift, and ReACT report traces
  • nao (agent-memory-system-review) - nao review: analytics-agent framework with file-system project context, SQL/runtime tools, skills, stories, evals, and chat-derived user memories
  • Napkin (agent-memory-system-review) - Napkin review: TypeScript/Bun markdown-vault CLI and SDK with progressive disclosure, Obsidian parity, BM25 cache, pi integration docs, and benchmark-only session mining
  • Nuggets (agent-memory-system-review) - Pi personal assistant with local HRR nugget files, prompt-time memory injection, chat scheduling, and partial promotion into Claude Code memory
  • o-o (agent-memory-system-review) - o-o review: polyglot HTML/bash living documents with embedded research contracts, source caches, freshness gates, index management, and Claude-only update dispatch
  • OpenSage (agent-memory-system-review) - OpenSage review: ADK agent framework with dynamic subagents, generated Skills, file memory, Neo4j graph memory, trace logging, plugins, and RL adapters
  • OpenViking (agent-memory-system-review) - OpenViking review: service-backed context database with viking:// filesystem URIs, AGFS/RAGFS storage, vector projections, session-derived memories, and agent plugins
  • Operational Ontology Framework (agent-memory-system-review) - Operational Ontology Framework review: public files-first artifact model for Pin/Spec/Facts/Handoff/Skills governance, with no published runner implementation
  • OS-Copilot (agent-memory-system-review) - OS-Copilot review: FRIDAY plans OS tasks, generates and repairs executable tools, scores them, and stores reusable Python tools in a vector-retrieved repository
  • Pal (agent-memory-system-review) - Agno personal knowledge agent with file/wiki compilation, PgVector routing metadata, Agno learnings, SQL facts, scheduled maintenance, and tool routing
  • Phantom (agent-memory-system-review) - Phantom review: AI co-worker runtime with prompt-block assembly, Qdrant memory, file-backed evolved identity, scheduler, MCP/UI surfaces, and invariant-checked reflection
  • Pi Self-Learning (agent-memory-system-review) - Pi extension that mines task-end session traces into mistake/fix reflections, scored core learnings, git-backed memory files, and prompt-time context
  • Playground (agent-memory-system-review) - Review of Playground, a TribleSpace-backed Rust agent runtime with shell-first turns, branch-separated traces, archive imports, and prompt-time memory covers
  • Pratiyush/llm-wiki (agent-memory-system-review) - Trace-derived local wiki builder that converts coding-agent session logs into raw markdown, synthesized wiki pages, static exports, MCP tools, and agent workflows
  • ReasoningBank (agent-memory-system-review) - Benchmark trace-to-memory system that extracts reasoning items from WebArena and SWE-Bench trajectories, retrieves by task-query embeddings, and reinjects selected items into prompts
  • Reflexion (agent-memory-system-review) - Reflexion review: benchmark trace-to-reflection loops that reinject failed-attempt plans into retries
  • REM (agent-memory-system-review) - REM review: database-backed episodic memory service that parses agent traces, retrieves prompt context, and partially wires LLM semantic consolidation
  • SAGE (agent-memory-system-review) - SAGE review: governed agent memory infrastructure with CometBFT writes, validator votes, RBAC, encryption, and MCP/REST recall surfaces
  • Self-Training-LLM (agent-memory-system-review) - Self-Training-LLM review: offline Wikipedia QA self-training pipeline that turns model answer rollouts and uncertainty scores into SFT/DPO datasets and checkpoints
  • Semiont (agent-memory-system-review) - Document-grounded annotation platform with W3C annotations, working-tree resources, git-staged event logs, shared human/agent write paths, and derived graph/vector projections
  • sift-kg (agent-memory-system-review) - LLM document-to-knowledge-graph pipeline with schema discovery, cached extractions, confidence/provenance fields, human review queues, narrative reports, graph viewers, and agent-facing query surfaces
  • Siftly (agent-memory-system-review) - Self-hosted X bookmark ingestion system that stores raw tweet/media evidence in SQLite, derives AI tags/categories, and serves hybrid search plus mindmap exploration
  • SignetAI (agent-memory-system-review) - SignetAI review: local-first cross-harness memory daemon with hooks, MCP, SQLite/FTS/vector/graph recall, raw transcripts, extraction gates, and procedural surfaces
  • SkillNote (agent-memory-system-review) - SkillNote review: self-hosted SKILL.md registry with Postgres-backed content, semver bundle releases, MCP/tool projection, agent sync plugins, and ratings without trace-derived learning
  • SkillWeaver (agent-memory-system-review) - SkillWeaver review: website-specific browser trajectories distilled into Playwright API skills, metadata counters, retrieval prompts, and executable tool authority
  • SkillX (agent-memory-system-review) - Trajectory-derived skill KB builder that distills successful task runs into plan, functional, and atomic prompt artifacts with partial retrieval support
  • Spacebot (agent-memory-system-review) - Rust concurrent agent framework with process-isolated channels, branches, workers, cortex synthesis, typed graph memory, and trace-derived persistence
  • Stash (agent-memory-system-review) - Stash review: self-hosted Postgres/pgvector MCP memory backend that consolidates agent-written episodes into facts, relationships, goals, failures, hypotheses, and decay state
  • Supermemory (agent-memory-system-review) - Supermemory review: hosted memory platform with open MCP, SDK middleware, graph UI contracts, and opaque hosted extraction/profile/ranking APIs
  • Synapptic (agent-memory-system-review) - Synapptic review: Claude Code trace compiler into weighted user profiles, behavioral guards, assistant memory files, benchmarks, and local relay index
  • Tendril (agent-memory-system-review) - Desktop Strands/Tauri agent sandbox that turns live task needs into workspace-local executable TypeScript capabilities
  • Thalo (agent-memory-system-review) - Thalo review: custom plain-text knowledge DSL with Tree-sitter grammar, entity schemas, semantic indexes, validation rules, LSP, Prettier, and synthesis actualization
  • Tolaria (agent-memory-system-review) - Tolaria review: Tauri desktop markdown/git vault with typed lenses, saved views, managed agent guidance, AI context snapshots, CLI-agent launchers, and MCP tools
  • Tracecraft (agent-memory-system-review) - Tracecraft review: S3/HuggingFace-backed CLI coordination layer with shared JSON state, messages, claims, barriers, and artifacts
  • Virtual Context (agent-memory-system-review) - Virtual Context review: proxy-owned context virtualization with durable conversation traces, compaction, facts, tag summaries, paging tools, and inspectable runtime surfaces
  • Voiden (agent-memory-system-review) - Voiden review: Git-native API workspace with executable .void documents, linked request-block reuse, local sidecars, request history, extensions, and agent skills
  • Voyager (agent-memory-system-review) - Embodied Minecraft agent that promotes critic-approved rollouts into retrievable JavaScript skills, with checkpointed traces, curriculum memory, QA cache, and Chroma indexes
  • WUPHF (agent-memory-system-review) - WUPHF review: local multi-agent office with fresh-session runners, git-backed wiki/notebooks, trace extraction, playbooks, skills, and linting
  • xMemory (agent-memory-system-review) - xMemory review: dialogue traces become episodes, semantic facts, theme summaries, kNN indexes, and entropy-gated adaptive retrieval for LoCoMo
  • Zikkaron (agent-memory-system-review) - Zikkaron review: Claude Code MCP memory server with SQLite, hooks, trace capture, hybrid retrieval, lifecycle compression, and generated instructions