Feature Reference
AI clients operate your workspace through MCP tools. Every tool is listed automatically inside the client (with parameter docs); this page is the capability overview.
Core tools
| Tool | Capability |
|---|---|
read | Single/multi-file reads (with version checks), directory listings |
edit | Line-level safe edits (Safe Edit), multi-file patches |
write | Create/overwrite whole files |
bash | One-shot commands and long tasks: paged output, stdin, cancel, wait |
Files and search
| Tool | Capability |
|---|---|
move_file | Move/rename files and directories; LSP updates code references |
read_image | Read image metadata or image content |
batch_file_tools | Run several read-only tools concurrently |
find_files | Glob file search |
search_files | ripgrep full-text search (rg ships with the installer) |
read_link | Fetch any URL and return raw HTML plus metadata |
Language intelligence (LSP)
Diagnostics and symbol tools run on language servers, routed by file extension with lazy startup:
| Tool | Capability |
|---|---|
get_diagnostics_code | File diagnostics (errors/warnings) |
search_symbols_code | Workspace symbol search |
get_symbol_definition_code | Hover/definition info |
get_document_symbols_code | Document outline |
Register servers per language in the config to enable them, e.g.:
"lsp": {
"servers": {
"typescript": { "command": "typescript-language-server", "args": ["--stdio"], "extensions": [".ts", ".tsx"] },
"python": { "command": "pyright-langserver", "args": ["--stdio"], "extensions": [".py"] },
"go": { "command": "gopls", "args": [], "extensions": [".go"] }
}
}Unconfigured languages return empty results — never an error.
Project memory
The memory tool group maintains per-workspace project memory:
memory_context: read prior experience at the start of a task (what was done, what broke);memory_record_summary: explicitly record a reusable summary at the end.
Memory is isolated per workspace UUID and stored locally; the panel can browse, delete or wipe it, and can disable it entirely.
Skills and agents
skills/*/SKILL.md and agents/*.md under the workspace (or resource root) are hot-registered as prompts under the same names — write your team's engineering conventions or deploy runbooks as markdown and the AI follows them. Edits take effect immediately.
The Safe Edit engine
The edit tool's line mode is not naive string replacement but a guard pipeline:
- Snapshot the file;
- Apply the edit;
- Risk analysis (bulk deletions, unbalanced braces, duplicate declarations — threshold hits are rejected and rolled back);
- Syntax validation (TS/JS via esbuild; built-in Vue SFC parsing);
- Automatic rollback on failure with a structured report (lines added/removed, per-check results).
Tool group switches
The panel's tool switches enable/disable tools per group (core / file / shell / diagnostics / symbol / search); changes hot-reload from the config without a restart.
Environment diagnostics
Run aeasymcp doctor for a health report: config file location, port usage, ripgrep availability, language server registrations — always the first step when something looks off.
Monitoring and logs
The panel provides per-tool call logs and statistics (success/failure/duration), resource status and error details; log files live under logs/ in the data directory.