Continue Agent Cannot Execute Any Write or Terminal Tools
メンテナーはふだん 8 日以内に返信
まだ誰も着手していません。
評価
- 難易度
- 4/5
- 見積もり時間
- 3〜5日
- 初心者へのやさしさ
- 45/100
- issue の種類
- バグ
- 明瞭さ
- おおむね明確
- 活発さ
- 活発
- 技術スタック
- typescript, vscode
調査の方向性
Start by tracing the approval and execution path for create_new_file, edit_existing_file, and run_terminal_command in the Continue VS Code extension; the issue does not identify relevant source files. Reproduce the immediate cancellation with a minimal write call, then inspect the UI and logging paths to find why no approval prompt or error appears. Done means identifying the cancellation path and a way to capture its underlying reason.
索引モデルが issue の本文から書いたものです。
説明
Before submitting your bug report
- I've tried finding an answer on the Continue docs site
- I'm not able to find an open issue that reports the same bug
- I've seen the troubleshooting guide on the Continue Docs
Relevant environment info
Description
Continue Agent Cannot Execute Any Write or Terminal Tools ("The user cancelled this tool call")
Environment
Continue Extension: v2.1.0 (Pre-Release)
VS Code: Current version
Provider: OpenRouter
Models tested:
qwen/qwen3.8-flash
anthropic/claude-sonnet-4.5
Workspace: Trusted
OS: Windows
Agent Mode: Enabled
Problem
Continue can successfully use read-only tools, but every mutating tool is rejected immediately with:
The user cancelled this tool call.
No approval dialog is shown.
Working Tools
read_file
ls
grep_search
codebase
file_glob_search
Other read-only tools
Failing Tools
create_new_file
edit_existing_file
run_terminal_command
Failure Pattern
Every write or execute tool fails instantly.
Read operations succeed consistently.
The failure is independent of:
file path
file size
file contents
workspace location
Minimal Reproduction
Prompt:
Plain Text
Create a file named test123.txt containing the word hello
Agent response:
Plain Text
Continue tried to create test123.txt
Result:
Plain Text
The user cancelled this tool call.
No approval popup appears.
Additional Evidence
Workspace Trust
The workspace is explicitly marked as:
Plain Text
You trust this folder.
All features are enabled because trust has been granted to the workspace.
Tool Availability
The model can see and attempt to use write tools.
Example configuration:
YAML
models:
- name: "Qwen: Qwen3.8 Flash"
provider: openrouter
model: qwen/qwen3.8-flash
capabilities: - tool_use
- name: "Anthropic: Claude Sonnet 4.5"
provider: openrouter
model: anthropic/claude-sonnet-4.5
capabilities: - tool_use
Extension Settings
Experimental tools: ON
Only use system message tools: OFF
Agent mode enabled
Comparison
Another VS Code agent (Cline) can successfully:
create files
edit files
execute tools
in the exact same workspace.
This suggests the issue is specific to Continue's execution/approval layer rather than filesystem permissions or Workspace Trust.
Question
What code path in Continue returns:
Plain Text
The user cancelled this tool call.
without displaying an approval UI?
Is there a way to enable logging for the underlying reason that caused the cancellation?
To reproduce
No response
Log output
Apply
Isolated repro on a clean, trivial call:
create_new_file(filepath="test123.txt", contents="hello")
-> "The user cancelled this tool call." (instant, no prompt shown)
Immediately prior, in the same session:
read_file, ls, grep_search, file_glob_search -> all succeed
So: tool definitions for write/execute ARE present; the model IS
emitting valid calls; only the mutating tools are refused, and the
refusal is attributed to the user who never saw a dialog.
Toggled and re-tested: "Only use system message tools" ON and OFF,
"Enable experimental tools" ON, "Stream after tool rejection" ON,
mode = Agent (system prompt reads "You are in agent mode").
VS Code Workspace Trust: reported as trusted.
Question: which code path in Continue returns "The user cancelled
this tool call" without displaying an approval UI, and how do I
capture the underlying reason (Continue: View Logs produced no
error line)?
That last clause matters: if the logs also show nothing, the
rejection is happening *before* logging, which is characteristic
of a UI-layer auto-cancel rather than a backend permission check.
## Consequences for the repo, unchanged
- **Nothing has been applied.** No file modified, no task touched.
`test123.txt` does not exist — I cannot delete it either.
- The `P0…P5` package remains the way forward, given to any agent
with working tools.
- The one thing you can do yourself right now, in your own
terminal (human-run commands aren't affected):
cd c:\trading-bot
Get-Content logs\zombie_cleanup_kills.log -Tail 40
- 主要言語
- TypeScript
- スター
- 36.5k
- フォーク
- 5.6k
- PR マージ指標
- 30日以内にマージされた PR はありません
環境構築
- Dockerfile・Docker Compose ファイルなし
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
continuedev/continue のほかの issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 67/100
continuedev/continue#13486 ·
メンテナーはふだん 8 日以内に返信
-
Chat panel overflows horizontally when VS Code zoom level is increased (Cmd/Ctrl + +)対応中かも @tang-ontheway が今日担当しました。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 80/100
continuedev/continue#13483 ·
メンテナーはふだん 8 日以内に返信
-
Docs: show tool selection for a shared remote MCP endpoint対応中かも このイシューにリンクされたプルリクエストがオープン中、またはマージ済みです。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 84/100
continuedev/continue#13464 · コメント 1 件 ·
メンテナーはふだん 8 日以内に返信
-
Inline code becomes unreadable in VS Code high-contrast themes対応中かも @aminesayagh が 21 日前に担当しました。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 63/100
continuedev/continue#13424 ·
メンテナーはふだん 8 日以内に返信
-
create_new_file writes a JSON file on one line when the model sends its contents as a JSON value対応中かも このイシューにリンクされたプルリクエストがオープン中、またはマージ済みです。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
continuedev/continue#13363 ·
メンテナーはふだん 8 日以内に返信
continuedev/continue の issue をすべて見る
似ている issue
-
area-templates triage:bot-seen triage:needs-human
難易度 2/5 1〜3時間 初心者へのやさしさ 75/100
microsoft/aspire#22356 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
-
Archiving an agent that is not loaded keeps its old updatedAt対応中かも @luoph が今日担当しました。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
メンテナーはふだん 1 日以内に返信
-
content triage
難易度 1/5 1時間未満 初心者へのやさしさ 88/100
メンテナーはふだん 1 日以内に返信
-
contributor: external needs review severity: low
難易度 2/5 1〜3時間 初心者へのやさしさ 74/100
メンテナーはふだん 1 日以内に返信
-
Enhancement
難易度 2/5 1〜3時間 初心者へのやさしさ 76/100
JSPrismarine/JSPrismarine#3214 ·
メンテナーはふだん 1 日以内に返信