Agent Loom
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| AGENT_LOOM_HTTP_TOKEN | Yes | The HTTP token that gates the MCP endpoint. Should be at least 24 random bytes. |
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
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| treeB | List files and directories inside the workspace, excluding blocked paths. |
| searchA | Use this for targeted verification or code lookup. Prefer one specific final search instead of repeated broad verification searches. |
| readA | Read a specific text file with line numbers. Avoid rereading files after write/edit/apply_patch unless exact final content is needed. |
| view_imageA | Inspect a PNG, JPEG, GIF, or WebP image from the active workspace. Returns native MCP image content plus dimensions and SHA-256. |
| writeA | Create or overwrite a meaningful text file inside the workspace. New files use an atomic rename; existing files retain their inode and metadata. Returns a unified diff; pass the SHA from read when overwriting shared files. |
| editA | Apply a targeted exact text replacement while retaining the existing file inode and metadata. Returns a unified diff; pass the SHA from read to reject stale multi-session edits. |
| apply_patchA | Apply one unified diff patch inside the workspace. Paths are validated before applying. Prefer edit for tiny replacements and apply_patch for multi-file diffs. |
| import_fileA | Import a ChatGPT Apps SDK attachment into the workspace. Accepts only a platform file object with download_url and file_id. Not a general URL downloader. Overwrite is off by default. |
| bashA | Run one allowlisted verification command in the workspace, such as tests, build, lint, typecheck, or a project script. Do not use for git status/diff or file inspection; use show_changes, tree, search, and read instead. Do not chain commands with &&, pipes, redirects, or shell file readers. |
| workspaceA | Route this chat to an allowed project. Use discover when a requested directory is not itself a Git repository; then open the relevant discovered repository before Git operations or starting Pi/Codex pools. Selection belongs only to this MCP session, so other chats may select different projects through the same endpoint. |
| piA | Unified persistent pi agent interface. For action=start, ALWAYS pass the explicit Git repository root from the user's request; MCP reconnects can reset implicit workspace selection. Start once, then send many tasks to stable h5i boxes and conversation sessions. Pools survive MCP requests and route back to their bound workspace by id. |
| codexA | Unified persistent codex agent interface. For action=start, ALWAYS pass the explicit Git repository root from the user's request; MCP reconnects can reset implicit workspace selection. Start once, then send many tasks to stable h5i boxes and conversation sessions. Pools survive MCP requests and route back to their bound workspace by id. |
| show_changesA | Summarize the current workspace changes in one review-oriented result with git status, diff stats, and optional diff. Use this instead of bash git status, bash git diff, git_status, or git_diff when reviewing work. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| agent-loom-tool-card | Compact visual renderer for Agent Loom workspace orientation, source changes, and agent operations. |
| agent-loom-tool-card-v9 | Compact visual renderer for Agent Loom workspace orientation, source changes, and agent operations. |
| agent-loom-tool-card-v8 | Compact visual renderer for Agent Loom workspace orientation, source changes, and agent operations. |
TDQS
Scored across 13 tools
Most tools are clearly separated by purpose: tree/search/read handle inspection, write/edit/apply_patch handle distinct modification styles, and bash/show_changes are explicitly kept apart from git/file review. The only mild overlap is between write, edit, and apply_patch, but their descriptions provide clear boundaries, and pi/codex are differentiated by agent name.
The tool names are readable and consistently lowercase/snake_case, but they mix bare verbs (write, read, edit, search), noun-like commands (tree, bash, workspace), and verb_noun forms (view_image, import_file, show_changes). This is a mixed convention rather than a predictable pattern.
Thirteen tools is a reasonable, well-scoped count for a coding workspace server. Each tool addresses a distinct operational need such as file editing, search, verification, change review, workspace routing, or persistent agent interaction, without feeling bloated.
The set covers core file operations, search, verification, change review, and persistent agent workflows well. However, there are notable gaps: no file delete/remove tool, and the workspace description references a discover capability that is not exposed in the tool list, which creates a potential dead end.