Jekyll2026-02-08T02:34:38+09:00https://rocknroll009.github.io/til/feed.xmlInf. 00’s TILThings I've learned.Multiple git account 깃 복수계정2026-02-04T00:00:00+09:002026-02-04T00:00:00+09:00https://rocknroll009.github.io/til/2026/02/04/multiple-gitThe cleanest ways to use multiple Git accounts locally are usually these two:

  1. Separate SSH keys per account (recommended — the most reliable)
  2. Use different user.name / user.email per repository (this only changes the commit author — push permissions/auth are separate)

Below is a standard setup for the case where both Account A and Account B are GitHub accounts, fully separated using SSH.

1) Generate SSH keys for each account

ssh-keygen -t ed25519 -C "accountA@email.com" -f ~/.ssh/id_ed25519_github_a
ssh-keygen -t ed25519 -C "accountB@email.com" -f ~/.ssh/id_ed25519_github_b

2) Create host aliases in ~/.ssh/config

Host github-a
  HostName github.com
  User git
  IdentityFile ~/.ssh/id_ed25519_github_a
  IdentitiesOnly yes

Host github-b
  HostName github.com
  User git
  IdentityFile ~/.ssh/id_ed25519_github_b
  IdentitiesOnly yes

3) Add the public keys to each GitHub account

Copy each public key and add it in GitHub → Settings → SSH and GPG keys.

cat ~/.ssh/id_ed25519_github_a.pub
cat ~/.ssh/id_ed25519_github_b.pub

4) Test the connection

ssh -T git@github-a
ssh -T git@github-b

If you see “Hi username!” for each one, it worked.

5) Decide which account the repo should use (clone/connect)

For example, if you want to push using Account B:

When cloning a new repo

git clone git@github-b:USERNAME/REPO.git

If the repo is already cloned, just change the remote URL

git remote set-url origin git@github-b:USERNAME/REPO.git

6) Separate commit author per repository (optional)

Inside the repo folder:

git config user.name "AccountB Name"
git config user.email "accountB@email.com"

Check:

git config user.name
git config user.email
git remote -v

Key takeaway for a GitHub Pages repo

  • Push authentication (permissions/login) is determined by whether origin uses git@github-a: or git@github-b:
  • The email shown in commits is determined by git config user.email
]]>openclaw local에서 test2026-02-04T00:00:00+09:002026-02-04T00:00:00+09:00https://rocknroll009.github.io/til/2026/02/04/openclaw
  • openclaw를 local에서 lmstudio 테스트하면서 TIL
    • # mac에서 docker start
colima start
docker info


# docker-compose symlink
mkdir -p ~/.docker/cli-plugins
ln -sf "$(which docker-compose)" ~/.docker/cli-plugins/docker-compose
docker compose version
brew install bash


# docker에서 테스트
# 결국 docker에서 local llm 돌리는 것은 포기
docker compose up -d openclaw-gateway
docker compose logs -f --tail=200 openclaw-gateway
docker compose up openclaw-gateway
docker compose run --rm openclaw-cli dashboard --no-open


# repo clone 사용
node dist/index.js onboard
node dist/index.jso
node dist/index.jso tui


# telegram 에러: docker에 openclaw 열어둬서 에러였음
For the Telegram 409 Conflict:
This usually means another instance is already connected. Check for running processes:
ps aux | grep clawdbot
ps aux | grep -i "openclaw\|dist/index.js" | grep -v grep

# reference
config: https://docs.openclaw.ai/gateway/configuration-examples
openclaw with vllm: https://unsloth.ai/docs/models/glm-4.7-flash
browser doc: https://docs.openclaw.ai/tools/browser
    • lm studio하려면 아래 .openclaw/openclaw.json 설정해야함
      "models": {
    "mode": "merge",
    "providers": {
      "lmstudio": {
        "baseUrl": "http://127.0.0.1:1234/v1",
        "apiKey": "lm-studio",
        "api": "openai-responses",
        "models": [
          {
            "id": "nvidia/nemotron-3-nano",
            "name": "lmstudio",
            "reasoning": false,
            "input": [
              "text"
            ],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "contextWindow": 16000,
            "maxTokens": 16000
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "lmstudio/nvidia/nemotron-3-nano"
      },
      "models": {
        "lmstudio/nvidia/nemotron-3-nano": {
          "alias": "Nemotron 3 Nano"
        }
      },
      "workspace": "/Users/user/.openclaw/workspace",
      "compaction": {
        "mode": "safeguard"
      },
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },
    ]]>    LangGraph 맛보기 feat. tool 디버깅2026-02-04T00:00:00+09:002026-02-04T00:00:00+09:00https://rocknroll009.github.io/til/2026/02/04/langgraphLangGraph 2가지 사용법
    • graph api: state, nodes and edges
    • functional api: persistence, memory, human-in-the-loop, and streaming
    • 대개 graph api 사용함.
    • functional api는 이미 다른 코드가 있고, 거기에 langgraph 접붙일 때, 코드 수정 최소화를 위해 사용

    • graph api Example code

      코드 보기 
      # Step 1: Define tools and model

from langchain.tools import tool
from langchain.chat_models import init_chat_model


model = init_chat_model(
    "claude-sonnet-4-5-20250929",
    temperature=0
)


# Define tools
@tool
def multiply(a: int, b: int) -> int:
    """Multiply `a` and `b`.

    Args:
        a: First int
        b: Second int
    """
    return a * b


@tool
def add(a: int, b: int) -> int:
    """Adds `a` and `b`.

    Args:
        a: First int
        b: Second int
    """
    return a + b


@tool
def divide(a: int, b: int) -> float:
    """Divide `a` and `b`.

    Args:
        a: First int
        b: Second int
    """
    return a / b


# Augment the LLM with tools
tools = [add, multiply, divide]
tools_by_name = {tool.name: tool for tool in tools}
model_with_tools = model.bind_tools(tools)

# Step 2: Define state

from langchain.messages import AnyMessage
from typing_extensions import TypedDict, Annotated
import operator


class MessagesState(TypedDict):
    messages: Annotated[list[AnyMessage], operator.add]
    llm_calls: int

# Step 3: Define model node
from langchain.messages import SystemMessage


def llm_call(state: dict):
    """LLM decides whether to call a tool or not"""

    return {
        "messages": [
            model_with_tools.invoke(
                [
                    SystemMessage(
                        content="You are a helpful assistant tasked with performing arithmetic on a set of inputs."
                    )
                ]
                + state["messages"]
            )
        ],
        "llm_calls": state.get('llm_calls', 0) + 1
    }


# Step 4: Define tool node

from langchain.messages import ToolMessage


def tool_node(state: dict):
    """Performs the tool call"""

    result = []
    for tool_call in state["messages"][-1].tool_calls:
        tool = tools_by_name[tool_call["name"]]
        observation = tool.invoke(tool_call["args"])
        result.append(ToolMessage(content=observation, tool_call_id=tool_call["id"]))
    return {"messages": result}

# Step 5: Define logic to determine whether to end

from typing import Literal
from langgraph.graph import StateGraph, START, END


# Conditional edge function to route to the tool node or end based upon whether the LLM made a tool call
def should_continue(state: MessagesState) -> Literal["tool_node", END]:
    """Decide if we should continue the loop or stop based upon whether the LLM made a tool call"""

    messages = state["messages"]
    last_message = messages[-1]

    # If the LLM makes a tool call, then perform an action
    if last_message.tool_calls:
        return "tool_node"

    # Otherwise, we stop (reply to the user)
    return END

# Step 6: Build agent

# Build workflow
agent_builder = StateGraph(MessagesState)

# Add nodes
agent_builder.add_node("llm_call", llm_call)
agent_builder.add_node("tool_node", tool_node)

# Add edges to connect nodes
agent_builder.add_edge(START, "llm_call")
agent_builder.add_conditional_edges(
    "llm_call",
    should_continue,
    ["tool_node", END]
)
agent_builder.add_edge("tool_node", "llm_call")

# Compile the agent
agent = agent_builder.compile()


from IPython.display import Image, display
# Show the agent
display(Image(agent.get_graph(xray=True).draw_mermaid_png()))

# Invoke
from langchain.messages import HumanMessage
messages = [HumanMessage(content="Add 3 and 4.")]
messages = agent.invoke({"messages": messages})
for m in messages["messages"]:
    m.pretty_print()

    tool call 탐구

    • ministral 모델 학습시 아마 아래와 같은 학습을 했을 것:
    [INST] What's 2+2? [/INST]
[TOOL_CALLS]  # <- 아래 log 보면 이 token이 생성되는 것을 확인할 수 있음
 [{"name": "calculator", "arguments": {...}}]
    • 위와 같은 tool use 학습하지 않은 모델은 agent 모델로 사용이 불가능함
    • tool call 할 때 실제 message body
    • lm studio 사용해서 디버깅해봄
    2026-02-08 01:13:16 [DEBUG]
 Received request: POST to /v1/chat/completions with body  {
  "messages": [
    {
      "content": "You are a helpful assistant",
      "role": "system"
    },
    {
      "content": "1*2",
      "role": "user"
    }
  ],
  "model": "mistralai/ministral-3-14b-reasoning",
  "stream": false,
  "temperature": 0,
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "add",
        "description": "Adds `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    },
    {
      "type": "function",
      "function": {
        "name": "multiply",
        "description": "Multiply `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    },
    {
      "type": "function",
      "function": {
        "name": "divide",
        "description": "Divide `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    }
  ]
}

# 모델의 tool call
2026-02-08 01:13:17 [DEBUG]
 [SamplingSwitch] Entering switch 'mistralV13ToolsSamplingSwitch' (triggered by string: "[TOOL_CALLS]") # <----------- tool call 토큰 생성함. tool use 학습의 산물.
2026-02-08 01:13:17  [INFO]
 [mistralai/ministral-3-14b-reasoning] Start to generate a tool call...
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Tool name generated:  multiply
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Accumulating tool call arguments:  {"
...
 [mistralai/ministral-3-14b-reasoning] Accumulating tool call arguments:  {"a": 1, "b": 2}
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Model generated a tool call.
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Model generated tool calls:  [multiply(a=1, b=2)]

# 모델 response
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Generated prediction:  {
  "id": "chatcmpl-37bu06198lu6b9atpcug9i",
  "object": "chat.completion",
  "created": 1770480796,
  "model": "mistralai/ministral-3-14b-reasoning",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "",
        "tool_calls": [ #<-------------- tool_call format에 맞춰서 response
          {
            "type": "function",
            "id": "570498487",
            "function": {
              "name": "multiply",
              "arguments": "{\"a\":1,\"b\":2}"
            }
          }
        ]
      },
      "logprobs": null,
      "finish_reason": "tool_calls"
    }
  ],
  "usage": {
    "prompt_tokens": 285,
    "completion_tokens": 17,
    "total_tokens": 302
  },
  "stats": {},
  "system_fingerprint": "mistralai/ministral-3-14b-reasoning"
}

# langgraph multiply tool의 결과값을 다시 llm으로 post
# function과 그 결과값을 돌려줌
2026-02-08 01:13:18 [DEBUG]
 Received request: POST to /v1/chat/completions with body  {
  "messages": [
    {
      "content": "You are a helpful assistant",
      "role": "system"
    },
    {
      "content": "1*2",
      "role": "user"
    },
    {
      "content": null,
      "role": "assistant",
      "tool_calls": [
        {
          "type": "function",
          "id": "570498487",
          "function": {
            "name": "multiply",
            "arguments": "{\"a\": 1, \"b\": 2}"
          }
        }
      ]
    },
    {
      "content": "2",
      "role": "tool",
      "tool_call_id": "570498487"
    }
  ],
  "model": "mistralai/ministral-3-14b-reasoning",
  "stream": false,
  "temperature": 0,
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "add",
        "description": "Adds `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    },
    {
      "type": "function",
      "function": {
        "name": "multiply",
        "description": "Multiply `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    },
    {
      "type": "function",
      "function": {
        "name": "divide",
        "description": "Divide `a` and `b`.\n\n    Args:\n        a: First int\n        b: Second int",
        "parameters": {
          "properties": {
            "a": {
              "type": "integer"
            },
            "b": {
              "type": "integer"
            }
          },
          "required": [
            "a",
            "b"
          ],
          "type": "object"
        }
      }
    }
  ]
}

# llm 모델의 최종 response
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Running chat completion on conversation with 4 messages.
2026-02-08 01:13:18  [INFO]
 [mistralai/ministral-3-14b-reasoning] Accumulated 1 tokens  The
...
2026-02-08 01:13:19  [INFO]
 [mistralai/ministral-3-14b-reasoning] Accumulated 14 tokens  The result of \(1 \times 2\) is **2**.
2026-02-08 01:13:19  [INFO]
 [mistralai/ministral-3-14b-reasoning] Model generated tool calls:  []
2026-02-08 01:13:19  [INFO]
 [mistralai/ministral-3-14b-reasoning] Generated prediction:  {
  "id": "chatcmpl-2ojj2bd99jrvma2y5bngh",
  "object": "chat.completion",
  "created": 1770480798,
  "model": "mistralai/ministral-3-14b-reasoning",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The result of \\(1 \\times 2\\) is **2**.",
        "tool_calls": []
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 305,
    "completion_tokens": 15,
    "total_tokens": 320
  },
  "stats": {},
  "system_fingerprint": "mistralai/ministral-3-14b-reasoning"
}

    reference

    • https://docs.langchain.com/oss/python/langgraph/quickstart
    • https://medium.com/ai-agents/langgraph-for-beginners-part-3-conditional-edges-16a3aaad9f31
    • https://github.com/teddylee777/LangGraph-Advanced-Tutorial/tree/main
    • https://langchain-ai.github.io/langgraph/tutorials/web-navigation/web_voyager/#agent-definition
    • https://github.com/spmallick/learnopencv/blob/master/LangGraph-A-Visual-Automation-and-Summarization-Pipeline/browser_automation.py
    • https://docs.mistral.ai/cookbooks/concept-deep-dive-tokenization-tool_calling
    ]]>