yoanbernabeu/grepai-skills
27 skills16.3k installs486 starsGitHub
Install
npx skills add https://github.com/yoanbernabeu/grepai-skillsSkills in this repo
1Grepai Search Basicsgrepai-search-basics is a grepai-skills entry covering how GrepAI semantic search differs from grep and ripgrep text matching. Prerequisites include running grepai init, creating an index with grepai watch, and having an embedding provider such as Ollama available. The skill walks basic search commands, result interpretation, and when meaning-based retrieval beats literal pattern search. Developers reach for grepai-search-basics when onboarding to GrepAI, exploring unfamiliar repositories, or wiring agents to query codebases by intent rather than exact identifiers.784installs2Grepai Ollama Setupgrepai-ollama-setup is a skill from yoanbernabeu/grepai-skills for installing and configuring Ollama as GrepAI's local embedding backend. It walks through first-time Ollama installation, embedding model selection, and connection troubleshooting so code never leaves the machine. The readme positions Ollama as enabling 100% private code search for GrepAI indexing workflows. Developers reach for this skill when setting up GrepAI with local embeddings, choosing models, or fixing Ollama connectivity during private repo search setup.726installs3Grepai Search Advancedgrepai-search-advanced is a Claude Code skill from yoanbernabeu/grepai-skills covering advanced GrepAI CLI options for AI agent integration. It documents --json or -j for machine-readable output, --toon or -t for TOON format, compact mode to shrink payloads, and --limit N with a default of 10 results for controlled context windows. Developers reach for this skill when wiring GrepAI into agent loops, CI scripts, or tooling that must process search hits programmatically while minimizing tokens sent to Claude or GPT models. Use it alongside base GrepAI search when plain-text ripgrep-style output is too verbose for multi-step agent reasoning or when downstream tools require schema-stable JSON or TOON records instead of prose summaries.706installs4Grepai Trace Graphgrepai-trace-graph is a developer skill for the GrepAI CLI that generates recursive call graphs and dependency trees from any entry function or method. It documents the `grepai trace graph` command, which walks callees recursively and renders a tree such as main → initialize → loadConfig → parseYAML for architecture and flow visualization. Developers reach for grepai-trace-graph when mapping complete function dependencies, understanding complex control flow, or scoping impact analysis before major refactors. The skill pairs with other grepai-skills for search and trace workflows and assumes grepai is installed in the environment.686installs5Grepai Trace Callersgrepai-trace-callers is a grepai-skills workflow for using the grepai trace callers command to answer who calls a given function or method. Developers use it before refactors, during impact analysis, and for code navigation when understanding dependencies matters. The skill documents when to trace callers—finding usages, assessing blast radius, and exploring call graphs—and pairs with GrepAI's CLI trace subsystem. Reach for grepai-trace-callers when renaming functions, changing signatures, or evaluating how widely a method is used across a codebase.671installs6Grepai Search Tipsgrepai-search-tips is a guidance skill for developers using GrepAI semantic search inside coding agents. The skill contrasts text grep with meaning-based retrieval, showing how to phrase intent-focused queries instead of exact symbol names—for example asking to retrieve a user from a database rather than searching for getUserById. It covers when to invoke the skill, a semantic search mindset table, and troubleshooting steps for poor result quality. Developers reach for grepai-search-tips when agent searches miss relevant files, return noisy matches, or when onboarding a team to GrepAI-backed workflows. The skill does not install GrepAI itself; it optimizes query writing so existing GrepAI integrations surface the right modules, configs, and call sites faster during exploration and refactors.667installs7Grepai Trace Calleesgrepai-trace-callees is a yoanbernabeu/grepai-skills package that teaches agents to run grepai trace callees against a named function and list all downstream calls. Given a function like ProcessOrder that invokes validateOrder, calculateTotal, and sendConfirmation, the command expands the full callee tree for dependency analysis. Developers reach for grepai-trace-callees when onboarding to unfamiliar modules, documenting behavior before refactors, or finding deeply nested dependencies without manual ripgrep chains. The skill answers the direct question what does this function call and pairs with other GrepAI trace skills for upstream caller analysis.651installs8Grepai Search Boostinggrepai-search-boosting is a skill from yoanbernabeu/grepai-skills for configuring score boosting in GrepAI agent code search. Boosting modifies search scores based on file paths—for example a +10% bonus raising a 0.85 src/auth.go hit to 0.935, or a -50% penalty dropping tests/auth_test.go to 0.425. Developers edit `.grepai/config.yaml` under `search.boost` to enable penalties for test, docs, and vendor patterns while boosting important directories. Reach for grepai-search-boosting when agent retrieval keeps surfacing test fixtures or third-party code instead of production source during implementation tasks.639installs9Grepai Embeddings Ollamagrepai-embeddings-ollama is an agent skill from yoanbernabeu/grepai-skills that configures Ollama as GrepAI's embedding backend for fully local, private code search. The workflow covers choosing an Ollama embedding model, wiring GrepAI to the local endpoint, and tuning performance while keeping source code on-machine with zero API cost. Developers reach for grepai-embeddings-ollama when corporate policy blocks cloud embeddings, air-gapped environments need offline search, or recurring embedding API fees are undesirable. The skill also documents troubleshooting for Ollama connection failures and latency tuning without network round-trips.629installs10Grepai Mcp Toolsgrepai-mcp-tools is a documentation skill from yoanbernabeu/grepai-skills that maps the GrepAI MCP server started with grepai mcp-serve. The reference covers tools such as grepai_search, which accepts a natural-language query string and an optional limit that defaults to 10 results, plus additional MCP endpoints described in the full skill file. Developers reach for grepai-mcp-tools when wiring GrepAI into an assistant, debugging malformed MCP calls, or learning embedding-based code search parameters before writing client configuration. The skill does not run searches itself; it teaches correct Model Context Protocol usage so agents invoke semantic retrieval instead of ripgrep alone. Use it alongside a running GrepAI index when onboarding agents to repository-aware lookup during implementation or code review.603installs11Grepai Chunkinggrepai-chunking is a Claude Code skill for configuring how GrepAI splits source files into embeddable segments before semantic search and indexing. The skill explains default ~512-token chunks, overlap tradeoffs, and adjustments for verbose versus concise code styles to improve retrieval accuracy. Developers reach for grepai-chunking when GrepAI search misses symbols, returns fragmented matches, or after onboarding a large monorepo to `.grepai/config.yaml`. It pairs with other GrepAI skills for languages and tracing when tuning a polyglot codebase index.596installs12Grepai Languagesgrepai-languages is a Claude Code skill documenting programming languages and file extensions GrepAI supports for semantic indexing and call-graph tracing. GrepAI indexes all text-based files but provides full Index+Trace support for Go (.go), JavaScript (.js, .jsx), TypeScript (.ts, .tsx), Python (.py), and PHP. Developers reach for grepai-languages before pointing GrepAI at a monorepo to verify trace capabilities, troubleshoot language-specific indexing gaps, or configure language-aware settings. The skill complements grepai-chunking when search issues stem from unsupported extensions rather than chunk boundaries.595installs13Grepai Watch Daemongrepai-watch-daemon configures and runs the GrepAI watch command for continuous codebase indexing. The daemon scans project source files, chunks code into roughly 512-token segments, generates embeddings through a configured provider, stores vectors in the configured backend, and monitors filesystem changes in the background. Developers reach for this skill when bootstrapping initial indexes, enabling real-time file watching, running the daemon detached, or troubleshooting stale semantic search results in GrepAI-powered agent workflows.590installs14Grepai Initgrepai-init is a setup skill from yoanbernabeu/grepai-skills that runs the grepai init command to create a .grepai directory in a project root. Initialization writes config.yaml for customization, prepares index.gob for the vector index populated by grepai watch, and symbols.gob for symbol metadata used in semantic search. Developers reach for grepai-init when onboarding GrepAI to a codebase for the first time or troubleshooting missing configuration before agent search sessions. The skill documents basic usage with cd into the project followed by grepai init, plus guidance on customizing initial settings and resolving initialization failures.589installs15Grepai Config Referencegrepai-config-reference is a yoanbernabeu/grepai-skills reference that explains all GrepAI settings stored at `/your/project/.grepai/config.yaml`, including the version field and advanced tuning blocks in the published schema excerpt. Developers reach for grepai-config-reference when optimizing GrepAI for a repo, troubleshooting misconfiguration, or setting up advanced search behavior before agents query the codebase. Use it whenever Claude needs authoritative option names and structure for GrepAI rather than guessing YAML keys during agent tooling setup or performance troubleshooting.584installs16Grepai Troubleshootinggrepai-troubleshooting is a diagnostic skill for GrepAI when semantic code search misbehaves on a developer machine. The guide supplies copy-paste commands such as grepai version, grepai status, curl against Ollama at localhost:11434/api/tags, and cat .grepai/config.yaml to inspect setup before deeper fixes. grepai-troubleshooting covers index-not-updating errors, poor search relevance, and connection or configuration failures common in local embedding pipelines. Developers reach for grepai-troubleshooting after GrepAI stops returning useful hits, embeddings fail to refresh, or Ollama connectivity breaks during agent-assisted codebase exploration. The skill structures quick diagnostics first, then walks through common issue sections with targeted remediation steps.582installs17Grepai Workspacesgrepai-workspaces is a yoanbernabeu/grepai-skills skill for configuring multi-project GrepAI workspaces. Workspaces let developers group monorepos, microservices, or related repos under one unified semantic search index with shared configuration. Commands like `grepai wo` create workspaces so agents and developers search all projects at once or filter to specific members. Use grepai-workspaces when onboarding to a multi-repo architecture, when watch indexing must cover sibling services, or when semantic code search should not stop at a single repository boundary.575installs18Grepai Mcp Claudegrepai-mcp-claude is a setup skill from yoanbernabeu/grepai-skills for connecting GrepAI to Claude Code through the Model Context Protocol. The GrepAI MCP server exposes semantic code search, call graph analysis, and index status monitoring to the assistant. Prerequisites include GrepAI installed, Ollama or another embedding provider running, the project indexed with grepai watch, and Claude Code installed. Developers reach for grepai-mcp-claude when enabling semantic search in Claude Code, configuring the MCP server, or troubleshooting Claude integration. One-command setup adds GrepAI tools so agents query meaningfully indexed code instead of raw ripgrep alone.571installs19Grepai Ignore Patternsgrepai-ignore-patterns is a configuration skill from yoanbernabeu/grepai-skills for controlling which paths GrepAI indexes. Ignore rules come from .grepai/config.yaml and are automatically merged with .gitignore patterns. Developers use grepai-ignore-patterns when excluding test files, generated code, vendor directories like node_modules, or other noise that bloats the index and dilutes search results. Pattern syntax supports exact directory names such as node_modules and vendor anywhere in the tree. Reach for this skill after initial grepai watch setup when search quality or index size needs tuning.567installs20Grepai Storage Gobgrepai-storage-gob is a configuration skill in yoanbernabeu/grepai-skills for enabling GrepAI's default GOB (Go Binary) file backend. GOB stores vector embeddings, file metadata, and chunk information in a single local file on disk, giving developers fast local semantic code search without PostgreSQL, Pinecone, or other external vector databases. The skill targets single-machine setups for small to medium codebases in local development environments where minimal dependencies matter. Developers reach for grepai-storage-gob when onboarding GrepAI on one workstation and want the simplest storage path before graduating to distributed backends.558installs21Grepai Mcp Cursorgrepai-mcp-cursor is an Agent Skill from yoanbernabeu/grepai-skills that configures GrepAI with Cursor using the Model Context Protocol. Cursor is an AI-powered IDE with MCP support; GrepAI adds semantic code search beyond plain text matching, call graph analysis for dependency tracing, and index-based navigation across large repositories. Developers reach for grepai-mcp-cursor when setting up GrepAI in Cursor, enabling semantic search for the IDE agent, configuring MCP endpoints, or troubleshooting integration failures. The skill covers setup and troubleshooting workflows so agents can find symbols and relationships by intent rather than exact string matches during refactors or onboarding.554installs22Grepai Quickstartgrepai-quickstart is a yoanbernabeu/grepai-skills walkthrough that gets GrepAI running on a local codebase in about five minutes. Prerequisites are terminal access and a code project to index. Step 1 installs GrepAI via Homebrew (brew install yoanbernabeu/tap/grepai) or a curl-based Linux/macOS script from the upstream repository. The skill targets first-time setup, quick refreshes, new-project indexing, and demos of semantic repository search to teammates. After install, agents gain locally indexed semantic search over the repo instead of relying on brittle text grep alone. Developers reach for grepai-quickstart when onboarding GrepAI before deeper grepai-skills workflows or when standing up agent context on a fresh checkout.550installs23Grepai Installationgrepai-installation is a multi-platform setup skill from yoanbernabeu/grepai-skills covering every method to install GrepAI on macOS, Linux, and Windows. The skill walks through Homebrew on macOS, package managers on Linux, PowerShell on Windows, and building from source with Go 1.24+. Developers use grepai-installation when installing GrepAI for the first time, upgrading an existing install, or confirming a successful setup via grepai version. Prerequisites include terminal access on Unix systems and PowerShell on Windows. The skill fits agent sessions where semantic grep capabilities are needed but the CLI is not yet present. Source builds require a Go toolchain; Homebrew is the recommended macOS path.548installs24Grepai Storage Qdrantgrepai-storage-qdrant is a yoanbernabeu/grepai-skills guide for using Qdrant as the GrepAI storage backend. Qdrant provides fast vector similarity search, scalability, advanced metadata filtering, and Docker-based deployment. The skill recommends Qdrant when search latency matters, codebases exceed 50K files, Qdrant infrastructure already exists, or advanced vector features are required. Prerequisites include a running Qdrant instance. Developers invoke grepai-storage-qdrant when GrepAI default storage bottlenecks on large monorepos and a production-grade vector engine is needed for agent-driven semantic code navigation.522installs25Grepai Embeddings Lmstudiogrepai-embeddings-lmstudio is a yoanbernabeu/grepai-skills agent skill for pointing GrepAI at LM Studio as the embedding backend. LM Studio provides a desktop GUI for downloading models and exposing an OpenAI-compatible embedding API, which GrepAI consumes for semantic code search without sending vectors to a cloud provider. The skill fits developers already running local LLMs in LM Studio who prefer visual model management over CLI-only embedding servers. Use it when configuring GrepAI for private embeddings, switching embedding models through a GUI, or integrating local inference with agent codebase retrieval. It complements other GrepAI provider skills by focusing specifically on LM Studio's OpenAI-compatible embedding endpoint rather than Ollama or cloud APIs.516installs26Grepai Storage Postgresgrepai-storage-postgres is a GrepAI storage-backend skill that configures PostgreSQL 14+ with the pgvector extension as the embedding index for semantic code search. Developers reach for it in team environments needing a shared index, codebases exceeding 10K files, concurrent indexer access, or reuse of existing PostgreSQL infrastructure. The skill covers database user permissions, network access to the server, and the tradeoffs versus local storage backends. Prerequisites include create-table permissions and reachable PostgreSQL with pgvector installed. Once configured, multiple developers or CI jobs can query and update the same vector index without rebuilding per-machine indexes.512installs27Grepai Embeddings Openaigrepai-embeddings-openai is a configuration skill for GrepAI that routes embedding generation through OpenAI's API instead of a self-hosted model. Developers use it when they want state-of-the-art vector quality, fast indexing without local GPU compute, and scalable search across large repositories. The skill documents trade-offs—privacy versus convenience—and when a shared team setup beats local inference. Reach for grepai-embeddings-openai when GrepAI is already in your workflow but local embeddings are too slow, too weak, or too costly to maintain.510installs