Documentation

Installation

The two paths are functionally identical — only the index backend differs. Requires Node >= 20.

The tarball prints a SQLITE BACKEND upgrade hint at startup; on a networked machine cd malong && npm ci upgrades it to the full backend.

Quick start

MCP clients

The MCP layer is standard JSON-RPC over stdio, so any MCP client works. All configs below are tested end-to-end.

{ "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "malong": {
      "type": "local",
      "command": ["node", "--max-old-space-size=512", "malong/mcp-server.js", "--workspace", "."],
      "enabled": true
    }
  }
}

{ "mcpServers": {
    "malong": { "command": "node", "args": ["/path/to/malong/mcp-server.js"] }
  }
}

claude mcp add liuhe -- node /path/to/malong/mcp-server.js --workspace /path/to/project
claude mcp list        # → "liuhe … ✔ Connected"

# Headless example
claude -p "index the workspace with reindex, then find createDb with symbol_search" \
  --allowedTools "mcp__liuhe__reindex" "mcp__liuhe__symbol_search"

model = "deepseek-v4-flash"
model_provider = "opencode-zen"

[model_providers.opencode-zen]
name = "OpenCode Zen"
base_url = "https://opencode.ai/zen/go/v1"   # any OpenAI-compatible endpoint
wire_api = "chat"
env_key = "OPENCODE_ZEN_API_KEY"             # codex reads the API key from this env var

[mcp_servers.liuhe]
command = "node"
args = ["/path/to/malong/mcp-server.js", "--workspace", "/path/to/project"]

Version note: recent codex releases force the OpenAI Responses API — use a version that still supports wire_api = "chat" (verified with 0.50.0).

bash malong/dsh/install-dsh.sh        # idempotent; edits ~/.dsh/profiles/web/cordis.patch.yml with backup
pkill -f "dsh web"; dsh web --port 3456 --host 0.0.0.0 --trusted-host <LAN IP>

The bridge exposes the Six-He line only — 13 tools as malong__<tool> — and auto-fills workspace_dir from the current conversation's workspace (explicit paths still win). Export MALONG_INTERFACE=classic|diagnostic on the dsh host for the classic full catalog (44) or the analyzer tier (18). Full guide incl. index rules: malong/dsh/DSH-INTEGRATION.md

CLI options

Environment variables

All optional — sensible defaults apply when unset.

13 MCP tools by default (Six-He line); classic 44 / diagnostic 18 on demand

Every tool is pure regex / AST — zero LLM calls, reproducible, auditable, CI-friendly. Expand a group for per-tool descriptions.

Tool Routing Map (classic 44 · appendix)

The default tier is the Six-He line (13 entries: code_task → code_read → code_patch / code_batch / code_maintain → code_check → code_control); see the home page for the full loop. The four sequence diagrams and the T01–T44 index below cover the classic 44 catalog and are kept as a deep reference. On the default tier, action-map rewrites classic tool names in responses into callable graph calls.

Expand: classic 44 full routing diagrams (4) + T01–T44 index

Legend

── ok = ok (green solid) = success → next step (every edge from measured handler next_step)

╌╌ err = err (red dashed) = failure → routing

thick green = main chain between layers

Node index (T01–T44) (44)

T01 reindex

reindex

ok → read_symbol / symbol_search / references

ok → health(check) self-check

T02 active

active_todos

err → high → address in current files

T03 ss

symbol_search

ok → impact_analysis

err → empty → code_search / code_query(mode=impact)

T04 cs

code_search

ok → read_symbol / references

T05 ft

find_tests

ok → test_bridge(scope)

err → none → create tests

T06 rs

read_symbol

ok → impact_analysis → edit_batch

after edit → test_bridge

T07 insp

inspect

ok → impact_analysis

after edit → test_bridge

T08 ol

read_outline

ok → impact_analysis

T09 cc

call_chain

ok → impact_analysis (full radius)

T10 refs

references

ok → find_tests (test refs)

T11 tr

trace_symbol

err → hardcoded copies → rename_symbol

T12 dg

dep_graph

err → cycles → fix_imports

ok → impact_analysis

T13 repo_map

read-only overview

T14 patch_parser

read-only parse preview

T15 ia

impact_analysis

err → high → sandbox_validate

err → medium → review callers

ok → edit · after → test_bridge

T16 ecg

edit_collision_guard

ok → safe → edit_transaction / edit_batch

err → external change → re-read → record_read

T17 et

edit_transaction

ok → commit → code_check(run=true) → code_query / invoke_capability

err → debug_runner

T18 eb

edit_batch

ok → test_bridge → debug_runner

via txn → diff_facts · TS → tsc_check

T19 ws

write_symbol

ok → test_bridge → debug_runner

via txn → diff_facts

T20 wss

write_symbols

ok → test_bridge → debug_runner

via txn → diff_facts

T21 rn

rename_symbol

dry_run → apply

ok → test_bridge

T22 sv

sandbox_validate

ok → edit_transaction

err → fix → re-validate

T23 gw

git_worktree

ok → commit → revert / reset (undo)

T24 ms

mock_sync

err → mismatch → fix → test_bridge

ok → test_bridge

T25 df

diff_facts

err → stale tests → test_bridge(scope)

err → callers → impact_analysis

ok → no sync issues

T26 tb

test_bridge

err → timeout → +timeout / inspect

err → fail → debug_runner / verify_pipeline

ok → pass → edit_transaction commit

T27 dr

debug_runner

err → suggested_action (14 types)

ok → continue

T28 tc

tsc_check

err → fix type errors

ok → continue

T29 vp

verify_pipeline

err → over-budget → single-stage lint

or test_bridge by scope

T30 cr

code_review

err → warnings → code_quality deep probe

after fix → test_bridge

T31 cq

code_quality

5-dim shape scores → review flagged

T32 sr

security_review

err → high → fix injection / secrets

ok → NOT a guarantee

T33 eg

exception_guard

err → issues → edit_transaction → test_bridge

test file → skip

T34 gp

guard_patterns

err → violations → fix → re-run

T35 nc

naming_consistency

err → issues → edit_transaction

T36 fi

fix_imports

err → issues → edit_transaction → test_bridge

ok → clean → sweep_dead_code

T37 sdc

sweep_dead_code

err → dead code → edit_transaction

unused_guard → manual trace

T38 cd

config_drift

err → drift → edit_transaction → re-run

ok → in sync

T39 dk

dependency_gatekeeper

err → missing → manifest → re-run

T40 sg

spec_gen

ok → read_symbol per export

T41 st

style_sniffer

ok → review rules → commit

T42 fb

feedback

err → issues → fix

ok → health(stats) usage

T43 hth

health

err → unregistered → register Y004 matrix

restart → MCP host · cleanup → re-run

T44 gcc

gc

memory reclaim (no next_step)

Language support

Symbol-level read / write (read & write_symbol) supports 10 language families, with automatic syntax self-check after writes (node --check / py_compile).

Safety mechanisms

Self-checks

FAQ