agent-code-intelligence

Plugin agent-code-intelligence from marketplace snowflake-ml-pipeline in vahid-ai/snowflake-ml-pipeline. Commands read from the repo's manifest — 1 month ago.

Skills (14)

• brief-readmes-inline-comments

  Keep README files brief and explain implementation beside the code when adding or updating source documentation, onboarding instructions, or READMEs.

• codebase-intelligence-router

  Route any codebase exploration, debugging, refactoring, architecture, dependency, documentation, or impact-analysis task to the cheapest and most precise available search/code-intelligence tool before reading large files or using expensive semantic/graph retrieval.

• docs-and-public-repos

  Retrieve current library/API documentation with Context7 and architecture/Q&A for unfamiliar public repositories with DeepWiki, then return to local code tools for precise implementation work.

• exact-search-cli

  Use fast deterministic CLI tools ripgrep, fd, jq, yq, and optionally fzf to locate exact code, files, errors, configuration, imports, or structured-data fields with minimal context.

• graph-impact-analysis

  Use CodeGraph or code-review-graph for blast radius, dependency/call paths, related tests, PR impact, and multi-hop structural reasoning after exact/LSP retrieval becomes insufficient.

• learn-from-failure

  Convert a debugging or tool failure into a durable lesson only after root cause and fix are verified, deduplicating against existing memory and recording evidence instead of speculative hypotheses.

• massive-code-search

  Use Zoekt indexed source search for very large monorepos or many repositories when recursive ripgrep would repeatedly scan too much data.

• mcp-profile-setup

  Generate conservative example MCP configurations for Serena, Context7, DeepWiki, Repomix, GrepAI, CodeGraph, Mem0, and optional memory/graph services without enabling every server by default.

• memory-router

  Decide where durable coding-agent knowledge should live: native Claude/Codex memory, local verified lessons, Mem0, Graphiti, Cognee, or Letta, while preventing redundant or unverified memory writes.

• repository-packing

  Use Repomix to create a bounded AI-friendly repository snapshot for one-shot review, model handoff, remote-repo analysis, or archival context, not as the default iterative coding retrieval mechanism.

• semantic-search

  Use GrepAI semantic search and call tracing only when the code concept is known but exact identifiers are not, or when semantic retrieval can replace broad exploratory file reading.

• structural-search

  Use ast-grep for syntax-aware code search and rewrites when the target is a program structure, invocation shape, declaration pattern, or codemod rather than plain text.

• symbol-navigation

  Navigate definitions, references, implementations, types, diagnostics, and call hierarchies using an existing LSP, JetBrains MCP, or Serena MCP instead of reading whole files.

• toolchain-doctor

  Check which code-intelligence CLIs and optional integration prerequisites are installed before choosing a retrieval backend or diagnosing missing MCP/LSP tools.

Subagents (1)

• codebase-intelligence

  Delegate proactively for any task that requires searching, navigating, exploring, or understanding a codebase. Use for locating implementations or tests, finding symbols and references, tracing call paths or dependencies, explaining architecture or unfamiliar code, identifying change locations, and

Hooks (1)

• PostToolUseFailure

  python3 ${PLUGIN_ROOT:-$CLAUDE_PLUGIN_ROOT}/scripts/capture_failure.py

Similar plugins