Skip to content

ai-coding-agent-workshop 残タスク #15

@littlemex

Description

@littlemex

Cline

  • remote ssh 試す

  • devbox 試す

  • PR: application inference profile

  • LiteLLM 経由で langfuse で可視化

  • q dev cli との使い分け

  • prompt cache with LiteLLM

  • cli, cdk, ssm plugin を自動インストールするスクリプト

  • web search mcp install

  • sonnet 3.7 failover が動かないので直す

  • top readme: 権限見直す

  • LiteLLM の詳細設定の反映

  • PR: LiteLLM 経由だと Application Inference Profile を設定できないので PR 出したい

  • PR: コード受け入れ率、コード生成数、を取りたい

  • LitelLM 経由だと $0 になっちゃうのでドルを裏で取りたい

  • docker logs が debug でログどんどん貯まるので info に変えてね

  • PR: コード行数が大量な時に開発がしづらい

  • aws サービスを正式名称にする

  • litellm 認証, https://docs.litellm.ai/docs/proxy/custom_auth, 認証時の情報をログに渡す

  • yaml と yml の表記ゆれ

  • .clinerules の配置と説明追加

  • langfuse の利用方法追加

  • LLMPerf

  • SSO 認証

  • https://github.com/aws-samples/genai-gateway、https://github.com/aws-samples/deploy-langfuse-on-ecs-with-fargate

  • litelllm のパススルー機能

  • cline rule に mcp を活用するように指示

  • aws mcp の kb 対応

  • web search, クレカ登録しなくても検索できる oss

  • langfuse sdk に cache_read_input_tokens 指定、callback で設定どうなるんだっけ

  • mlflow span 関連エラーの問題

litellm-1   | 2025/04/18 17:30:41 DEBUG mlflow.tracing.utils: Failed to get attribute mlflow.experimentId with from span _Span(name="litellm-acompletion", context=SpanContext(trace_id=0xe6cbe5b4e9802359f4aba821e6443add, span_id=0xecf2926cba77bf4c, trace_flags=0x01, trace_state=[], is_remote=False)).
litellm-1   | Traceback (most recent call last):
litellm-1   |   File "/usr/lib/python3.13/site-packages/mlflow/tracing/utils/__init__.py", line 201, in get_otel_attribute
litellm-1   |     return json.loads(span.attributes.get(key))
litellm-1   |            ~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm-1   |   File "/usr/lib/python3.13/json/__init__.py", line 339, in loads
litellm-1   |     raise TypeError(f'the JSON object must be str, bytes or bytearray, '
litellm-1   |                     f'not {s.__class__.__name__}')
litellm-1   | TypeError: the JSON object must be str, bytes or bytearray, not NoneType
  • mlflow 設定ファイル

Managed MLflow を利用できるようにした上で Prompt Caching を有効化し、Role 認証する Config の作成が今後必要です。
mlflow_callback.py で boto3 をアクセスキーで利用しているため、Role 認証の設定を試す必要があります。

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions