{"id":43116515,"url":"https://github.com/rendro/sediment","last_synced_at":"2026-02-14T03:22:58.418Z","repository":{"id":335109723,"uuid":"1143804293","full_name":"rendro/sediment","owner":"rendro","description":"Semantic memory for AI agents. Local-first, MCP-native.","archived":false,"fork":false,"pushed_at":"2026-02-07T03:20:06.000Z","size":253,"stargazers_count":0,"open_issues_count":1,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-07T12:30:55.691Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://sediment.sh/","language":"Rust","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/rendro.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-01-28T01:48:55.000Z","updated_at":"2026-02-07T01:06:29.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/rendro/sediment","commit_stats":null,"previous_names":["rendro/sediment"],"tags_count":15,"template":false,"template_full_name":null,"purl":"pkg:github/rendro/sediment","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rendro%2Fsediment","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rendro%2Fsediment/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rendro%2Fsediment/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rendro%2Fsediment/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/rendro","download_url":"https://codeload.github.com/rendro/sediment/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rendro%2Fsediment/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29433304,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-14T02:20:56.896Z","status":"ssl_error","status_checked_at":"2026-02-14T02:11:29.478Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-01-31T19:04:23.011Z","updated_at":"2026-02-14T03:22:58.399Z","avatar_url":"https://github.com/rendro.png","language":"Rust","funding_links":[],"categories":[],"sub_categories":[],"readme":"[![Crates.io](https://img.shields.io/crates/v/sediment-mcp.svg)](https://crates.io/crates/sediment-mcp)\n[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)\n[![CI](https://github.com/rendro/sediment/actions/workflows/ci.yml/badge.svg)](https://github.com/rendro/sediment/actions/workflows/ci.yml)\n\n# Sediment\n\nSemantic memory for AI agents. Local-first, MCP-native.\n\nCombines vector search, a relationship graph, and access tracking into a unified memory intelligence layer — all running locally as a single binary.\n\n## Why Sediment?\n\n- **Single binary, zero config** — no Docker, no Postgres, no Qdrant. Just `sediment`.\n- **Sub-16ms recall** — local embeddings and vector search at 100 items, no network round-trips.\n- **4-tool focused API** — `store`, `recall`, `list`, `forget`. That's it.\n- **Works everywhere** — macOS (Intel + ARM), Linux x86_64. All data stays on your machine.\n\n### Comparison\n\n| | Sediment | OpenMemory MCP | mcp-memory-service |\n|---|---|---|---|\n| Install | Single binary | Docker + Postgres + Qdrant | Python + pip |\n| Dependencies | None | 3 services | Python runtime + deps |\n| Tools | 4 | 10+ | 24 |\n| Embeddings | Local (all-MiniLM-L6-v2) | API-dependent | API-dependent |\n| Graph features | Built-in | No | No |\n| Memory decay | Built-in | No | No |\n\n## Install\n\n```bash\n# Via crates.io\ncargo install sediment-mcp\n\n# Via Homebrew\nbrew tap rendro/tap\nbrew install sediment\n\n# Via shell installer\ncurl -fsSL https://raw.githubusercontent.com/rendro/sediment/main/install.sh | sh\n\n# From source\ncargo install --path .\n```\n\n## Setup\n\nAdd Sediment to your MCP client configuration:\n\n### Claude Desktop\n\nAdd to `~/Library/Application Support/Claude/claude_desktop_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"sediment\": {\n      \"command\": \"sediment\"\n    }\n  }\n}\n```\n\n### Claude Code\n\nRun `sediment init` in your project, or add manually to `~/.claude/settings.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"sediment\": {\n      \"command\": \"sediment\"\n    }\n  }\n}\n```\n\n### Cursor\n\nAdd to `.cursor/mcp.json` in your project:\n\n```json\n{\n  \"mcpServers\": {\n    \"sediment\": {\n      \"command\": \"sediment\"\n    }\n  }\n}\n```\n\n### VS Code (Copilot)\n\nAdd to `.vscode/mcp.json` in your project:\n\n```json\n{\n  \"servers\": {\n    \"sediment\": {\n      \"command\": \"sediment\"\n    }\n  }\n}\n```\n\n### Windsurf\n\nAdd to `~/.codeium/windsurf/mcp_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"sediment\": {\n      \"command\": \"sediment\"\n    }\n  }\n}\n```\n\n### JetBrains IDEs\n\nGo to **Settings \u003e Tools \u003e AI Assistant \u003e MCP Servers**, click **+**, and add:\n\n```json\n{\n  \"sediment\": {\n    \"command\": \"sediment\"\n  }\n}\n```\n\n## Tools\n\n| Tool | Parameters | Description |\n|------|------------|-------------|\n| `store` | `content`, `scope?` | Save content to memory |\n| `recall` | `query`, `limit?` | Search by semantic similarity |\n| `list` | `limit?`, `scope?` | List stored items |\n| `forget` | `id` | Delete an item by ID |\n\n## CLI\n\n```bash\nsediment           # Start MCP server\nsediment init      # Set up Claude Code integration\nsediment stats     # Show database statistics\nsediment list      # List stored items\n```\n\n## How It Works\n\n### Two-Database Hybrid\n\nAll local, embedded, zero config:\n\n- **LanceDB** — Vector embeddings and semantic similarity search\n- **SQLite** (`access.db`) — Relationship graph, access tracking, decay scoring, consolidation queue\n\n### Key Features\n\n- **Memory decay**: Results re-ranked by freshness (30-day half-life) and access frequency. Old memories rank lower but are never auto-deleted.\n- **Trust-weighted scoring**: Validated and well-connected memories score higher.\n- **Project scoping**: Automatic context isolation between projects. Same-project items get a similarity boost.\n- **Relationship graph**: Items linked via RELATED, SUPERSEDES, and CO_ACCESSED edges. Recall expands results with 1-hop graph neighbors and co-access suggestions.\n- **Background consolidation**: Near-duplicates (≥0.95 similarity) auto-merged; similar items (0.85–0.95) linked.\n- **Type-aware chunking**: Intelligent splitting for markdown, code, JSON, YAML, and plain text.\n- **Conflict detection**: Items with ≥0.85 similarity flagged on store.\n- **Cross-project recall**: Results from other projects flagged.\n- **Local embeddings**: all-MiniLM-L6-v2 via Candle (384-dim vectors, no API keys).\n- **Model integrity**: SHA-256 verification of all model files on every load, pinned to a specific revision.\n- **Auto-migration**: Database schema automatically migrated when upgrading from older versions.\n\n### Security\n\n- **Input bounds**: Content (1MB), queries (10KB), JSON-RPC lines (10MB).\n- **Rate limiting**: 600 tool calls per minute.\n- **SQL injection prevention**: Sanitized filter expressions for LanceDB; parameterized queries for SQLite.\n- **Cross-project access control**: Forget enforces project isolation. Cross-project content is flagged in recall results.\n- **Error sanitization**: Internal errors logged to stderr; only generic messages returned to MCP clients.\n- **Retry with backoff**: Transient failures retried with exponential backoff (3 attempts, 100ms–2s).\n\n## Performance\n\nSub-16ms recall at 100 items with full graph features enabled. See [BENCHMARKS.md](BENCHMARKS.md) for detailed numbers.\n\n| DB Size | Graph Off | Graph On |\n|---------|-----------|----------|\n| 100     | ~12ms     | ~15ms    |\n| 1,000   | ~36ms     | ~65ms   |\n\n### Data Location\n\n- Vector store: `~/.sediment/data/`\n- Graph + access tracking: `~/.sediment/access.db`\n\nEverything runs locally. Your data never leaves your machine.\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for build instructions and PR guidelines.\n\n## License\n\n[MIT](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frendro%2Fsediment","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Frendro%2Fsediment","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frendro%2Fsediment/lists"}