CICADA

mcp-name: io.github.wende/cicada

Code Intelligence: Contextual Analysis, Discovery, and Attribution

Context compaction for AI code assistants – Give your AI structured, token-efficient access to 17+ languages including Elixir, Python, TypeScript, JavaScript, Rust, and more.

Up to 50% less waiting · Up to 70% less tokens · Up to 99% less explanations to do Tighter context = Better Quality

Quick Install · Security · Developers · AI Assistants · Docs

Why CICADA?

The core problem: AI code assistants waste context on blind searches. Grep dumps entire files when you only need a function signature, leaving less room for actual reasoning.

The Context Compaction Approach

Instead of raw text dumps, CICADA gives your AI structured, pre-indexed knowledge:

What You Get

• AST-level indexing – Module/function/class definitions with signatures, specs, docs
• 17+ language support – Elixir, Python, TypeScript, JavaScript, Rust, Go, Java, Kotlin, Scala, C/C++, Ruby, C#, Visual Basic, Dart, PHP, Erlang (beta)
• Complete call-site tracking – Aliases, imports, dynamic references across all supported languages
• Semantic search – Find code by concept with keyword extraction or embeddings (Ollama integration)
• Git + PR attribution – Surface why code exists, not just what
• Dependency analysis – Bidirectional tracking (what calls this, what does this call)
• Automatic language detection – Works seamlessly across polyglot codebases

Install

# 1. Install uv (if needed)
# curl -LsSf https://astral.sh/uv/install.sh | sh
uv tool install cicada-mcp

# In your repo
cicada claude   # or: cicada cursor, cicada vs, cicada gemini, cicada codex, cicada opencode, cicada zed

uvx cicada-mcp claude   # or cursor, vs

claude mcp add cicada uvx cicada-mcp

gemini mcp add cicada uvx cicada-mcp

codex mcp add cicada uvx cicada-mcp

kimi mcp add --transport stdio cicada -- cicada-mcp

Available commands after installation:

• cicada [claude|cursor|vs|gemini|codex|opencode|zed] - One-command interactive setup per project
• cicada-mcp - MCP server (auto-started by editor)
• cicada serve - Start REST API server for HTTP access to all MCP tools
• cicada status - Show index status, PR index, link status, agent files, MCP configs
• cicada stats [repo] - Display usage statistics (tool calls, tokens, execution times)
• cicada watch - Watch for file changes and automatically reindex
• cicada index - Re-index code with custom options (-f/--force, --keywords, --embeddings, --watch)
• cicada index-pr - Index pull requests for PR attribution
• cicada run [tool] - Execute any of the 7 MCP tools directly from CLI
• cicada agents install - Install Claude Code agents to ./.claude/ directory
• cicada link [parent_dir] - Links current repository to an existing index
• cicada clean - Completely removes cicada integration from your folder as well as all settings

Ask your assistant:

# Elixir
"Show me the functions in MyApp.User"
"Where is authenticate/2 called?"

# Python
"Show me the AuthService class methods"
"Where is login() used in the codebase?"

# Both languages
"Find code related to API authentication"

Privacy & Security

• 100% local: parsing + indexing happen on your machine; no external access.
• No telemetry: CICADA doesn't collect usage or any telemetry.
• Read-only tools: MCP endpoints only read the index; they can't change your repo.
• Optional GitHub access: PR features rely on gh and your existing OAuth token.
• Data layout:

  ~/.cicada/projects/<repo_hash>/
  ├─ index.json      # modules, functions, call sites, metadata
  ├─ config.yaml     # indexing options + mode
  ├─ hashes.json     # incremental indexing cache
  └─ pr_index.json   # optional PR metadata + reviews
  

  Your repo only gains an editor config (.mcp.json, .cursor/mcp.json, .vscode/settings.json, .gemini/settings.json, .codex/mcp.json, or .opencode.json).

For Developers

Wire CICADA into your editor once, and every assistant session inherits the context.

Install & Configure

cd /path/to/project
cicada claude   # or cicada cursor / cicada vs / cicada gemini / cicada codex / cicada opencode / cicada zed

Enable PR Attribution (optional)

brew install gh    # or apt install gh
gh auth login
cicada index-pr .     # incremental
cicada index-pr . --clean   # full rebuild

Unlocks questions like "Which PR introduced line 42?" or "What did reviewers say about billing.ex?"

Automatic Re-indexing with Watch Mode

Enable automatic reindexing when files change by starting the MCP server with the --watch flag:

** .mcp.json**

{
  "mcpServers": {
    "cicada": {
      "command": "cicada-mcp",
      "args": ["--watch"],
      "env": {
        "CICADA_CONFIG_DIR": "/home/user/.cicada/projects/<hash>"
      }
    }
  }
}

When watch mode is enabled:

• A separate process monitors .ex, .exs (Elixir) and .py (Python) files for changes
• Changes are automatically reindexed (incremental, fast)
• 2-second debounce prevents excessive reindexing during rapid edits
• The watch process stops automatically when the MCP server stops
• Excluded directories: deps, _build, node_modules, .git, assets, priv, .venv, venv

CLI Cheat Sheet

Note: Language detection is automatic – CICADA detects Elixir (mix.exs) and Python (pyproject.toml) projects automatically.

Troubleshooting

cicada index /path/to/project

cicada index .

# Install GitHub CLI
brew install gh  # macOS
sudo apt install gh  # Ubuntu
# or visit https://cli.github.com/

# Authenticate
gh auth login

# Index PRs
cicada index-pr

cicada index-pr --clean

# Re-index with keyword extraction
cicada index .  # or --keywords

cat ~/.cicada/projects/<hash>/config.yaml
# Should show:
# indexing:
#   mode: keywords

More detail: PR Indexing, Incremental Indexing.

# Ensure .venv is excluded
echo "/.venv/" >> .gitignore

# Use keywords mode for quickest indexing
cicada index --keywords .

For AI Assistants

CICADA ships 7 focused MCP tools designed for efficient code exploration across Elixir, Python, and Erlang codebases.

🧭 Which Tool Should You Use?

Want to see these tools in action? Check out Complete Workflow Examples with pro tips and real-world scenarios.

Core Tools

query - Smart code discovery (your starting point)

• Automatically detects keywords vs patterns
• Filters: scope (public/private), recent (last 14 days), filter_type (modules/functions), match_source (docs/strings)
• Returns snippets with smart next-step suggestions
• Use path_pattern to filter by location

search_module - Deep module analysis

• View complete API: functions, signatures, specs, docs
• For Python: Shows classes with method counts and signatures
• For Elixir: Shows functions with arity notation
• Bidirectional analysis:
  • what_calls_it=true → See who uses this module (impact analysis)
  • what_it_calls=true → See what this module depends on
• Supports wildcards (Elixir: MyApp.*, Python: api.handlers.*) and OR patterns (MyApp.User|MyApp.Post)
• Filter by visibility (public/private/all)

search_function - Function usage tracking

• Find definitions and all call sites
• what_calls_it=true (default) → See all callers
• what_it_calls=true → See all dependencies
• Include code examples with include_usage_examples=true
• Filter by usage_type: source, tests, or all

Git History (Unified Tool)

git_history - All git operations in one tool

• Single line: git_history("file.ex", start_line=42) → blame + PR
• Line range: git_history("file.ex", start_line=40, end_line=60) → grouped blame
• Function tracking: git_history("file.ex", function_name="create_user") → evolution
• File history: git_history("file.ex") → all PRs/commits
• Time filtering: recent=true (14d), recent=false (>14d), recent=null (all)
• Author filtering: author="john"
• Automatic PR index integration when available

Additional Tools

expand_result - Drill down from query results

• Auto-detects module vs function
• Shows complete details with usage examples
• Configure what to include: code, dependencies, callers
• Convenient wrapper around search_module and search_function

query_jq - Advanced index queries

• Direct jq queries against the index
• Schema discovery with | schema
• Compact (default) or pretty output
• Sample mode for large results

Detailed parameters + output formats: MCP_TOOLS_REFERENCE.md.

Token-Friendly Responses

All tools return structured Markdown/JSON snippets (signatures, call sites, PR metadata) instead of full files, keeping prompts lean.

New in v0.5.1: All tools now use compact output by default to minimize token usage. Use verbose=true for detailed output with full docs and specs.

Documentation

• Codebook – Complete feature reference and user guides
• Workflows – Real-world examples chaining tools together
• Installation – Step-by-step setup for all editors
• Contributing – Development guidelines and architecture
• CHANGELOG.md – Release notes

Deep Dives:

• Keyword Extraction Analysis – Semantic search internals
• PR Indexing – GitHub integration details
• MCP Tool Call Benchmarking – Token/time benchmarks
• Tool Discoverability – UX improvements research

Roadmap

Current Status

Production Ready:

• ✅ Elixir (tree-sitter)
• ✅ Python (SCIP)
• ✅ TypeScript (SCIP)
• ✅ JavaScript (SCIP)
• ✅ Rust (SCIP)

Beta:

• 🚧 Erlang (tree-sitter)
• 🚧 Go (SCIP)
• 🚧 Java/Kotlin/Scala (SCIP)
• 🚧 C/C++ (SCIP)
• 🚧 Ruby (SCIP)
• 🚧 C#/Visual Basic (SCIP)
• 🚧 Dart (SCIP)
• 🚧 PHP (SCIP)

Comparison to Alternatives

When to choose CICADA: You want local-first operation with rich git context (PR attribution, blame, function evolution tracking) and efficient token usage.

When to choose Serena: You need code editing capabilities through LSP and can accept higher resource usage.

When to choose Codicil: You have an Elixir project and prefer LLM-powered semantic summaries (Elixir-only).

Contributing

git clone https://github.com/wende/cicada.git
cd cicada
uv sync
pytest

Before submitting a PR:

• Run black cicada tests
• Ensure tests + coverage pass (pytest --cov=cicada --cov-report=term-missing)
• Update docs if behaviour changes

We welcome issues/PRs for:

• New language grammars
• Tool output improvements
• Better onboarding docs and tutorials

License

MIT – see LICENSE.