Trust & Safety

Logician provides granular control over what the agent can do, from read-only planning to fully autonomous editing.

Permission modes

Mode	Description	File edits	Tool execution
plan	Read-only, suggests changes	❌	❌
ask	Asks before each action	✅ (confirm)	✅ (confirm)
acceptEdits	Edits automatically, asks for tools	✅	⚠️ (confirm)
acceptAll	Full autonomy	✅	✅

Safe edits

All file edits use strict exact-text matching:

1. Read first — the agent reads the file before editing
2. Exact match — edits use the exact text from the file
3. CRLF preservation — line endings are preserved
4. BOM handling — byte order marks are handled correctly
5. Path normalization — paths are normalized before operations

// Edit uses exact text from file
oldText: "function login(user) { return false; }"
newText: "function login(user) { return validate(user); }"

Tool restrictions

Control which tools the agent can use:

{
  "tools": {
    "allowed": ["read_file", "grep", "find"],
    "denied": ["bash", "edit_file"]
  }
}

File access control

Restrict which directories the agent can access:

{
  "access": {
    "allowedPaths": ["src/", "tests/"],
    "deniedPaths": ["node_modules/", ".git/"]
  }
}

Audit trail

Every action is logged:

[10:23:45] read_file: src/auth.ts
[10:23:46] edit_file: src/auth.ts (3 changes)
[10:23:47] bash: npm test (exit 0)
[10:23:52] write_file: docs/changelog.md