chatgpt-codex-local-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MCP_HOST | No | Host to bind the HTTP server (if using HTTP transport). | 127.0.0.1 |
| MCP_PORT | No | Port to bind the HTTP server. | 3333 |
| MCP_LOG_LEVEL | No | Log level (info, debug, etc.). | info |
| MCP_TRANSPORT | No | Transport mode: stdio or http. | stdio |
| MCP_ENABLE_WRITE | No | Enable write operations (false by default). | false |
| MCP_ALLOWED_ROOTS | No | Comma-separated list of allowed root directories. If not set, uses current directory. | |
| MCP_ENABLE_NETWORK | No | Enable network access (false by default). | false |
| MCP_MAX_FILE_BYTES | No | Maximum file size in bytes to read. | 200000 |
| MCP_ENABLE_COMMANDS | No | Enable project commands from allowlist (false by default). | false |
| MCP_MAX_OUTPUT_BYTES | No | Maximum output bytes from commands. | 120000 |
| MCP_COMMAND_TIMEOUT_MS | No | Timeout in milliseconds for command execution. | 120000 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| list_allowed_repositoriesA | List configured allowed roots and likely project directories under them. |
| detect_project_stackC | Detect common stack markers for a project directory. |
| list_project_filesA | List files under a project directory inside an allowed root. |
| read_fileA | Read a text file inside an allowed root. Sensitive files are blocked. |
| search_in_projectB | Search for literal text in text files under a project directory. |
| get_git_statusA | Run a read-only git status for a project directory inside an allowed root. |
| get_git_diffA | Run a read-only git diff for a project directory inside an allowed root. |
| read_memory_indexA | Read docs/memory/index.md for navigation and current status. |
| read_onboarding_memoryA | Read docs/memory/onboarding.md to bootstrap a new project session. |
| read_project_memoryB | Read docs/memory/projeto.md with functional project knowledge. |
| read_architecture_memoryB | Read docs/memory/arquitetura.md with technical architecture context. |
| read_decisions_memoryA | Read docs/memory/decisoes.md with simplified ADR records. |
| read_backlog_memoryA | Read docs/memory/backlog.md with TODO, DOING, and DONE items. |
| read_pending_memoryA | Read docs/memory/pendencias.md with open issues and risks. |
| read_chat_context_memoryA | Read docs/memory/contexto-chat.md with current operational session context. |
| read_references_memoryA | Read docs/memory/referencias.md with important links and references. |
| read_important_files_memoryB | Read docs/memory/arquivos-importantes.md with key project files. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 17 tools
Each tool targets a distinct resource/action. The general exploration tools (list, read, search, git) are clearly separated, and each memory tool reads a specific, uniquely named file with no overlap.
Tool names follow a consistent verb-first pattern: list_*, read_*, search_*, get_*. The memory tools all use read_<topic>_memory, with read_memory_index as a minor but still predictable deviation. Overall naming is highly uniform.
At 17 tools, the server is slightly above the ideal 3-15 range. The 9 memory read tools are nearly identical operations on different files, inflating the count and adding unnecessary redundancy. A generic read_memory_file would reduce this.
For a read-only project exploration and memory reading server, the tool surface covers the core workflows: listing files, reading content, searching, and git status/diff. Minor gaps include no git log or branch listing, and memory files not in the predefined set require using read_file.