Code intelligence MCP server with a Zig core for local project indexing, structural outlines, symbol lookup, search, dependency graphs, snapshots, remote public-repo queries, and fallback edits.
CodeDB indexes local projects and exposes file tree, outline, search, symbol, caller, dependency, read, snapshot, project, and context tools to the MCP client., CodeDB's `codedb_edit` tool exists as a fallback editing tool and can create, replace, insert, delete, or modify files when used by a client without native edit tooling., The npm package runs a postinstall step that downloads a native binary from GitHub Releases; review package and release provenance in environments that restrict native binaries., Remote repo queries use the public `api.wiki.codes` service and should be treated as network access outside the local repository., The upstream README marks the project as alpha software, with parser coverage and snapshot formats still stabilizing.
Privacy notes
Local indexes, snapshots, file trees, symbol names, dependency graphs, snippets, read results, and search results can reveal proprietary code structure and implementation details., Upstream documents sensitive-file blocking for patterns such as environment files, credentials, and keys, but users should still review ignore rules and avoid indexing secret-heavy directories., CodeDB writes telemetry to `~/.codedb/telemetry.ndjson` unless `CODEDB_NO_TELEMETRY=1` is set, then syncs aggregate tool counts, latency, startup, file count, line count, language, version, and platform data on MCP session close., Upstream telemetry docs state that source code, file contents, file paths, and search queries are not collected., Remote public-repo queries and local MCP responses may still be logged by MCP clients, model providers, and terminal history.
Author
justrach
Submitted by
oktofeesh1
Claim status
unclaimed
Last verified
2026-06-06
Decision playbook
Review trust signals before you adopt
Signals are present but mixed. Use the checklist below to confirm the source and operational safety for your environment.
Compare context
Selected
0
Current score
63
Baseline
—
Delta
No baseline selected
No major trust-signal divergence detected in the current selection.
Source and provenance checks
Needs review
Confirm ownership and provenance before trusting install instructions.
Source link availableRequired
Open the canonical repository and verify ownership.
Done
Source provenance statusRequired
Marked as source-backed.
Done
Metadata reviewed
No reviewed flag detected in metadata.
Pending
Safety and privacy checks
Complete
Validate risk disclosures before installation or API wiring.
Safety notes presentRequired
Review the listed safety guidance before running commands.
Done
Privacy notes presentRequired
Review data handling notes before connecting accounts or secrets.
Done
Trust level risk gateRequired
Trust level does not block evaluation.
Done
Package and install checks
Needs review
Check package metadata and artifact integrity signals.
Install payload available
Install or copy payload is available for review.
Done
Package verification flag
No package verification flag provided.
Pending
Checksum metadata
No checksum provided for downloaded artifact.
Pending
Compare-driven decision checks
Needs review
Use compare context to validate trade-offs before adoption.
Compare tray has multiple entries
Add at least one more entry to compare trust differences.
5 safety and 5 privacy notes across 4 risk areas. Review closely: credentials & tokens, network access.
4 areas
SafetyLocal filesCodeDB indexes local projects and exposes file tree, outline, search, symbol, caller, dependency, read, snapshot, project, and context tools to the MCP client.
SafetyLocal filesCodeDB's `codedb_edit` tool exists as a fallback editing tool and can create, replace, insert, delete, or modify files when used by a client without native edit tooling.
SafetyNetwork accessThe npm package runs a postinstall step that downloads a native binary from GitHub Releases; review package and release provenance in environments that restrict native binaries.
SafetyNetwork accessRemote repo queries use the public `api.wiki.codes` service and should be treated as network access outside the local repository.
SafetyGeneralThe upstream README marks the project as alpha software, with parser coverage and snapshot formats still stabilizing.
PrivacyLocal filesLocal indexes, snapshots, file trees, symbol names, dependency graphs, snippets, read results, and search results can reveal proprietary code structure and implementation details.
PrivacyCredentials & tokensUpstream documents sensitive-file blocking for patterns such as environment files, credentials, and keys, but users should still review ignore rules and avoid indexing secret-heavy directories.
PrivacyCredentials & tokensCodeDB writes telemetry to `~/.codedb/telemetry.ndjson` unless `CODEDB_NO_TELEMETRY=1` is set, then syncs aggregate tool counts, latency, startup, file count, line count, language, version, and platform data on MCP session close.
PrivacyLocal filesUpstream telemetry docs state that source code, file contents, file paths, and search queries are not collected.
PrivacyNetwork accessRemote public-repo queries and local MCP responses may still be logged by MCP clients, model providers, and terminal history.
Disclosure: Alpha code intelligence MCP server. The npm package is named `codedeebee` because the bare `codedb` name is restricted on npm, and it installs a native `codedb` binary launcher.
Safety notes
CodeDB indexes local projects and exposes file tree, outline, search, symbol, caller, dependency, read, snapshot, project, and context tools to the MCP client.
CodeDB's `codedb_edit` tool exists as a fallback editing tool and can create, replace, insert, delete, or modify files when used by a client without native edit tooling.
The npm package runs a postinstall step that downloads a native binary from GitHub Releases; review package and release provenance in environments that restrict native binaries.
Remote repo queries use the public `api.wiki.codes` service and should be treated as network access outside the local repository.
The upstream README marks the project as alpha software, with parser coverage and snapshot formats still stabilizing.
Privacy notes
Local indexes, snapshots, file trees, symbol names, dependency graphs, snippets, read results, and search results can reveal proprietary code structure and implementation details.
Upstream documents sensitive-file blocking for patterns such as environment files, credentials, and keys, but users should still review ignore rules and avoid indexing secret-heavy directories.
CodeDB writes telemetry to `~/.codedb/telemetry.ndjson` unless `CODEDB_NO_TELEMETRY=1` is set, then syncs aggregate tool counts, latency, startup, file count, line count, language, version, and platform data on MCP session close.
Upstream telemetry docs state that source code, file contents, file paths, and search queries are not collected.
Remote public-repo queries and local MCP responses may still be logged by MCP clients, model providers, and terminal history.
Prerequisites
macOS or Linux on x64 or arm64 for the published native binary launcher.
Node.js 18 or newer when using the `codedeebee` npm launcher.
A local project directory that the MCP client exposes through roots or launches from.
Review of which local repositories, file types, generated artifacts, and secrets patterns may be indexed.
Optional `CODEDB_NO_TELEMETRY=1` if local aggregate telemetry should not be synced upstream.
Schema details
Install type
cli
Troubleshooting
No
Source repository stats
Scope
Source repo
Collection metadata
Estimated setup
10 minutes
Difficulty
intermediate
Tool listing metadata
Disclosure
Alpha code intelligence MCP server. The npm package is named `codedeebee` because the bare `codedb` name is restricted on npm, and it installs a native `codedb` binary launcher.
CodeDB is a code intelligence MCP server with a Zig core. It indexes local
projects and exposes MCP tools for file trees, outlines, symbol lookup,
trigram-accelerated search, word lookup, callers, dependency graphs, context
composition, hot files, reads, snapshots, remote public-repo queries, and
fallback edits.
Use it when Claude needs fast structural code context without reading entire
files first. CodeDB is designed as a context engine for coding agents: it helps
an agent find, understand, and retrieve relevant code while leaving most edits
to the client's native editing tools.
These sources were reviewed on 2026-06-06. Prefer the live repository, MCP
setup guide, npm registry metadata, license, README, npm launcher manifest,
launcher script, MCP implementation, fallback edit implementation, and
telemetry documentation for current installation and behavior details.
Features
Index local project structure and maintain project snapshots.
Inspect trees, symbols, outlines, imports, callers, dependencies, changed
files, hot files, and project status.
Search with trigram and word indexes.
Read file ranges with compact output and hash guards.
Compose task-shaped context from natural-language requests.
Query indexed public GitHub repositories through the remote tool.
List locally indexed projects.
Use a fallback edit tool for clients that do not provide native file editing.
Run as a stdio MCP server through the native codedb binary or the
codedeebee npm launcher.
Installation
For MCP clients that can run npx, use the npm launcher:
The upstream docs also describe a one-line installer and direct binary setup.
Review the installer and native-binary download path before using those options
in locked-down environments.
Use Cases
Ask Claude to locate a symbol definition before editing related code.
Retrieve outlines and dependencies instead of loading whole files.
Search a large repository quickly after indexing.
Build task-specific context from a natural-language coding request.
Inspect callers and dependency paths for a refactor.
Query an indexed public GitHub repository without cloning it locally.
Use fallback edits only when the MCP client has no native edit tool.
Safety and Privacy
CodeDB is powerful because it indexes local code. Point it only at repositories
that are appropriate for the MCP client and model provider to inspect. Review
generated snapshots, ignored paths, and sensitive-file handling before indexing
private workspaces.
The codedeebee npm package downloads a native codedb binary during
postinstall. Organizations with supply-chain controls should verify package,
release, and checksum behavior before installing it automatically.
Telemetry is enabled by default according to upstream docs, writes locally to
~/.codedb/telemetry.ndjson, and syncs aggregate usage and performance data on
MCP session close unless CODEDB_NO_TELEMETRY=1 is set. Review this behavior
before using CodeDB with confidential or regulated repositories.
Duplicate Check
Existing entries cover Serena, Context Mode, CodeGraphContext, LeanCTX,
language servers, repository search, and other code intelligence tools, but no
CodeDB entry, justrach/codedb, codedeebee package, or matching source URL
was found in content/mcp.
Show that CodeDB MCP Server is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/mcp/codedb-mcp-server)
How it compares
CodeDB MCP Server side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
Code intelligence MCP server with a Zig core for local project indexing, structural outlines, symbol lookup, search, dependency graphs, snapshots, remote public-repo queries, and fallback edits.
Official Agentset MCP server that lets Claude retrieve cited knowledge-base results from an Agentset namespace through the `knowledge-base-retrieve` tool, with optional tenant scoping and custom tool descriptions.
Local code indexing MCP server that lets AI assistants search repositories, build symbol indexes, inspect file summaries, monitor changes, and navigate multi-language codebases.
Apache-licensed TypeScript MCP server for Obsidian vaults with stdio and Streamable HTTP transports, Local REST API access, note search, section-aware edits, tag/frontmatter management, path policies, and read-only controls.
✓CodeDB indexes local projects and exposes file tree, outline, search, symbol, caller, dependency, read, snapshot, project, and context tools to the MCP client.
CodeDB's `codedb_edit` tool exists as a fallback editing tool and can create, replace, insert, delete, or modify files when used by a client without native edit tooling.
The npm package runs a postinstall step that downloads a native binary from GitHub Releases; review package and release provenance in environments that restrict native binaries.
Remote repo queries use the public `api.wiki.codes` service and should be treated as network access outside the local repository.
The upstream README marks the project as alpha software, with parser coverage and snapshot formats still stabilizing.
✓The MCP server sends Claude's retrieval queries to the Agentset API using the configured API key and namespace.
The `knowledge-base-retrieve` tool can return up to 100 results per call and can rerank results by relevance.
Namespace and tenant selection control which indexed documents are searchable; review them before connecting a shared agent.
API keys should be scoped, rotated, and stored only in the MCP server environment or a secret manager.
Custom tool descriptions can influence when the model calls the retrieval tool, so review them before use in production workflows.
✓Code Index MCP reads repository files, builds shallow and deep indexes, exposes file summaries, and can return source snippets through search and file resources.
Point `--project-path` at one approved repository; avoid broad parent directories, home directories, generated dependency trees, or secret-heavy paths.
Deep indexing and watcher refreshes can consume CPU, memory, disk IO, and temp storage on large codebases.
The server can create and clear index/cache settings and temporary directories used for stored index data.
Search results, file summaries, and symbol bodies can surface credentials, private code, customer data, internal URLs, or proprietary implementation details if they exist in the repository.
✓Cyanheads Obsidian MCP can read, list, search, create, append, patch, replace, delete, tag, and edit frontmatter in the configured vault through the Local REST API plugin.
The server defaults to full-vault access unless read paths, write paths, or read-only mode are configured.
Set `OBSIDIAN_READ_ONLY=true` for research-only workflows, then narrow writes with `OBSIDIAN_WRITE_PATHS` before allowing note edits.
Delete operations are permanent in the vault and should require human review even when the MCP host supports destructive-tool prompts.
Command-palette tools are disabled unless `OBSIDIAN_ENABLE_COMMANDS=true`; command execution can trigger opaque or destructive Obsidian plugin behavior.
Streamable HTTP deployments should use JWT or OAuth auth before being exposed beyond a trusted local machine.
Privacy notes
✓Local indexes, snapshots, file trees, symbol names, dependency graphs, snippets, read results, and search results can reveal proprietary code structure and implementation details.
Upstream documents sensitive-file blocking for patterns such as environment files, credentials, and keys, but users should still review ignore rules and avoid indexing secret-heavy directories.
CodeDB writes telemetry to `~/.codedb/telemetry.ndjson` unless `CODEDB_NO_TELEMETRY=1` is set, then syncs aggregate tool counts, latency, startup, file count, line count, language, version, and platform data on MCP session close.
Upstream telemetry docs state that source code, file contents, file paths, and search queries are not collected.
Remote public-repo queries and local MCP responses may still be logged by MCP clients, model providers, and terminal history.
✓Retrieved chunks can include private documents, product specs, policies, support content, internal procedures, historical project information, or customer-specific data.
Retrieval queries, namespace IDs, tenant IDs, document chunks, citations, and tool outputs may be visible to the MCP client, model provider, Agentset logs, and application telemetry.
Tenant IDs are useful for data segregation, but incorrect tenant or namespace configuration can expose the wrong knowledge base.
Do not paste API keys, namespace IDs, tenant IDs, or retrieved private chunks into shared issue reports, screenshots, or repository files.
✓Tool calls may expose file paths, filenames, source code snippets, imports, classes, methods, functions, symbols, search matches, complexity summaries, and project settings.
Index data and temp directories can persist derived repository structure or symbol metadata outside the source tree.
File watcher status, exclude patterns, and settings can reveal local project layout and development conventions.
If an MCP client forwards tool results to a model provider, repository contents and search snippets may leave the local machine.
Exclude secrets, generated artifacts, vendored dependencies, build output, and private data before indexing.
✓Vault notes, headings, block references, frontmatter, tags, file metadata, outgoing links, search queries, command names, plugin status, and edit history may be exposed to the MCP client and model provider.
API keys, local plugin endpoints, HTTP auth settings, path allowlists, logs, OpenTelemetry traces, and Docker environment variables can reveal private vault structure or credentials.
Tag listing is vault-wide in upstream behavior and can reveal tag names outside narrowed read paths.
Notes may contain personal journals, client work, credentials, meeting notes, research plans, health data, or other sensitive knowledge-base content.
Prerequisites
macOS or Linux on x64 or arm64 for the published native binary launcher.
Node.js 18 or newer when using the `codedeebee` npm launcher.
A local project directory that the MCP client exposes through roots or launches from.
Review of which local repositories, file types, generated artifacts, and secrets patterns may be indexed.
Agentset account or self-hosted Agentset deployment with a populated namespace.
Agentset API key with access to the namespace Claude should query.
Node.js 18.17 or newer for running the `@agentset/mcp` package.
Namespace ID selected with `--ns` or `AGENTSET_NAMESPACE_ID`.
Python 3.10 or newer.
uv or another Python package runner that can install `code-index-mcp`.
An approved repository path that the MCP client is allowed to inspect.
Optional native search tools such as ripgrep, ugrep, ag, or grep for faster search backends.
Node.js 24 or newer, or Bun 1.3.11 or newer.
Obsidian vault with the Obsidian Local REST API community plugin enabled.
Local REST API key stored in the MCP client environment.
Stdio MCP client, or reviewed Streamable HTTP deployment with authentication.
Install
npx -y codedeebee mcp
Run `npx @agentset/mcp --ns <namespace-id>` with `AGENTSET_API_KEY` set in the MCP server environment.