feat(security): fill gaps vs Claude Code#23883
Open
excess122 wants to merge 12 commits intoanomalyco:devfrom
Open
feat(security): fill gaps vs Claude Code#23883excess122 wants to merge 12 commits intoanomalyco:devfrom
excess122 wants to merge 12 commits intoanomalyco:devfrom
Conversation
- bash.ts: add DANGEROUS_PATTERNS static check (25 rules) - system.ts: add GitHub Copilot GPT-5 model routing - file/protected.ts: add Linux sensitive dir protection - worktree/index.ts: use Shell.preferred() on Windows
github-actions
Bot
added
needs:compliance
Contributor
|
Thanks for updating your PR! It now meets our contributing guidelines. 👍 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue for this PR
Closes #
Type of change
What does this PR do?
Fills four functional gaps compared to Claude Code's implementation:
** ool/bash.ts — Dangerous command static check**: Added DANGEROUS_PATTERNS array (25 rules) and checkDangerous() function that intercepts high-risk commands before execution (e.g.,
m -rf /, disk format with mkfs/ disk, boot record wipe with dd if=..., history deletion). Inspired by Claude Code's hardcoded blacklist of ~23 dangerous patterns. The check runs synchronously after command parsing, before the permission prompt.
session/system.ts — GitHub Copilot GPT-5 routing: Added a routing branch at the top of the provider() function so that when both providerID contains github-copilot and �pi.id contains gpt-5, the dedicated copilot-gpt-5.txt system prompt is used instead of the default one. The prompt file already existed but was never wired up.
** ile/protected.ts — Linux sensitive directory protection**: Added LINUX_HOME constant covering ~/.ssh, ~/.gnupg, ~/.aws, ~/.config/gcloud, ~/.kube, ~/.docker, ~/.netrc. Previously,
ames() and paths() returned empty sets on Linux, leaving sensitive credential and key directories unprotected from file scanning.
worktree/index.ts — Windows shell selection: Replaced hardcoded ["cmd", ["/c", cmd]] with Shell.preferred() so worktree start commands use pwsh/powershell when available on Windows, consistent with how ool/bash.ts selects the shell.
I understand why each change works: (1) pattern matching before execution prevents the dangerous commands from running at all; (2) the routing check mirrors the existing �nthropic/openai routing logic; (3) Linux was simply missing from a platform switch; (4) Shell.preferred() already handles the pwsh vs cmd decision tree.
How did you verify your code works?
Screenshots / recordings
N/A — no UI changes.
Checklist