{"data":{"slug":"minishlab-semble","name":"semble","tagline":"Fast and Accurate Code Search for Agents","github_url":"https://github.com/MinishLab/semble","owner":"MinishLab","repo":"semble","owner_avatar_url":"https://avatars.githubusercontent.com/u/177965497?v=4","primary_language":"Python","stars":5927,"forks":256,"topics":["agents","code-search","embeddings","mcp","mcp-server","model-context-protocol","retrieval"],"archived":false,"github_pushed_at":"2026-08-12T05:30:37+00:00","maintenance_label":"Active","stars_delta_30d":247,"url":"https://www.graphcanon.com/tools/minishlab-semble","markdown_url":"https://www.graphcanon.com/tools/minishlab-semble.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/minishlab-semble","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=minishlab-semble","description":"Fast and Accurate Code Search for Agents. Uses 99% fewer tokens than grep+read","homepage_url":"https://minish.ai/packages/semble/introduction/","license":"MIT","open_issues":3,"watchers":18,"ai_summary":"A tool that enables efficient code search using significantly fewer tokens compared to grep+read methods.","readme_excerpt":"

\n \"semble
\n Fast and Accurate Code Search for Agents
\n Uses ~99% fewer tokens than grep+read\n

\n\n
\n

\n \"Package\n \n \"Codecov\"\n \n \n \"License\n \n

\n\n[Quickstart](#quickstart) •\n[CLI](#cli) •\n[MCP Server](#mcp-server) •\n[Installation](docs/installation.md) •\n[Benchmarks](#benchmarks)\n\n\n
\n\nSemble is a code search library built for agents. It returns the exact code snippets they need instantly, using ~99% fewer tokens than grep+read. Indexing and searching a full codebase end-to-end takes under a second, matching the retrieval quality of a code-specialized transformer while indexing ~220x faster and querying ~17x faster (see [benchmarks](#benchmarks)). Everything runs on CPU with no API keys, GPU, or external services. Use it as an MCP server, a CLI tool via AGENTS.md, or a dedicated sub-agent, and any coding agent (Claude Code, Cursor, Codex, OpenCode, etc.) gets instant access to any repo.\n\n## Quickstart\n\nYour agent queries Semble in natural language (e.g. `\"How is authentication handled?\"`) and gets back only the relevant code snippets, without grepping or reading full files.\n\nThe fastest way to get started is the interactive installer. Install [uv](https://docs.astral.sh/uv/getting-started/installation/), then run:\n\n```bash\nuv tool install semble\nsemble install\n```\n\n`semble install` detects installed coding agents such as Claude Code, Codex, and OpenCode, and then lets you choose which integrations to enable:\n\n- **MCP server**: lets the agent call Semble directly as a tool.\n- **Instructions**: adds CLI usage guidance to AGENTS.md / CLAUDE.md.\n- **Sub-agent**: installs a dedicated `semble-search` sub-agent.\n\nTo undo the setup, run `semble uninstall`.\n\nFor manual setup instructions (MCP config per agent, AGENTS.md snippet, sub-agent files), see the [installation docs](docs/installation.md).\n\n
\nUpdating Semble\n\n```bash\nuv tool upgrade semble # upgrade\nuv cache clean semble # for MCP users (restart your MCP client after)\n```\n\n
\n\n
\nUnattended install\n\nFor sandboxed or scripted environments, skip the prompts with `--agent` and, optionally, `--type`:\n\n```bash\nsemble install --agent claude --type mcp subagent --yes\n```\n\n`--agent` accepts one or more agent ids (e.g. `claude`, `codex`, `pi`); `--type` accepts `mcp`, `instructions`, `subagent`, or `all` (default: all); `--yes` skips the confirmation prompt (requires `--agent` for a fully non-interactive run).\n\n
\n\n## Main Features\n\n- **Fast**: indexes an average repo in ~500 ms and answers queries in ~1 ms, all on CPU.\n- **Accurate**: NDCG@10 of 0.854 on our [benchmarks](#benchmarks), on par with code-specialized transformer models, at a fraction of the size and cost.\n- **Token-efficient**: returns only the relevant chunks, using [~99% fewer tokens than grep+read](#benchmarks).\n- **Zero setup**: runs on CPU with no API keys, GPU, or external services required.\n- **MCP server**: works with Claude Code, Cursor, Codex, OpenCode, VS Code, and any other MCP-compatible agent.\n- **Local and remote**: pass a local path or a git URL.\n\n## CLI\n\nSemble also ships as a standalone CLI. This is useful in scripts or anywhere you want search results without an MCP session. Indexes are built and cached on first run, and invalidated automatically when files change.\n\n```bash\n# Search a local repo","github_created_at":"2026-04-06T14:56:45+00:00","created_at":"2026-07-11T11:28:41.151245+00:00","updated_at":"2026-08-22T06:00:46.0472+00:00","categories":[{"slug":"ai-agents","name":"AI Agents","url":"https://www.graphcanon.com/categories/ai-agents","markdown_url":"https://www.graphcanon.com/categories/ai-agents.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/ai-agents"},{"slug":"data-retrieval","name":"Data & Retrieval","url":"https://www.graphcanon.com/categories/data-retrieval","markdown_url":"https://www.graphcanon.com/categories/data-retrieval.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/data-retrieval"}],"tags":[{"slug":"agents","name":"agents"},{"slug":"code-search","name":"code-search"},{"slug":"embeddings","name":"embeddings"},{"slug":"mcp","name":"mcp"},{"slug":"retrieval","name":"retrieval"}],"trust":{"provenance":{"is_fork":false,"github_id":1202937420,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-22T06:00:45.244Z","maintenance":{"label":"Active","score":82,"methodology":"github_public_v1","releases_90d":14,"days_since_push":10,"last_release_at":"2026-08-12T05:30:37Z","stars_delta_30d":247,"open_issues_delta_30d":-4},"security_summary":{"status":"no_manifest","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T11:28:42.327Z","medium_count":0,"scan_profile":"mcp_manifest","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-08-22T06:00:45.709Z"},"has_cli":{"value":true,"source":"pyproject.toml:[project.scripts]","observed_at":"2026-08-22T06:00:45.709Z"},"languages":{"value":["python"],"source":"github.language+pyproject.toml","observed_at":"2026-08-22T06:00:45.709Z"},"license_spdx":{"value":"MIT","source":"github.license","observed_at":"2026-08-22T06:00:45.709Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":{"notes":["Operating with Python, Semble does not require Docker for its operation."],"requires_docker":false},"constraints":{"requires_docker":false},"when_to_use":["- Use Semble when you are specifically working with AI agents or models and require efficient, token-economical code search operations.","- It is ideal when your application involves heavy reliance on embeddings for rapid retrieval of relevant code snippets."],"when_not_to_use":["- Avoid using Semble if your use case does not involve AI agents or the model-context-protocol (MCP). Competitor tools might offer better features tailored to non-agent-based code search.","- Not recommended in scenarios where token efficiency is not a concern, as competitors may provide more versatile functionalities without focusing on token reduction."],"source":"enrich:decision_facts","observed_at":"2026-07-12T06:32:26.336Z"},"constraint_facets":{"requires_docker":false},"decision_summary":[{"label":"Requirements","value":"Operating with Python, Semble does not require Docker for its operation."},{"label":"Adopt for","value":"Semble is a Python-based tool that facilitates fast and accurate code search for AI agents with up to 98% fewer tokens compared to traditional grep+read methods."}]}}