test: add unit coverage for permissions module

[Ayush7614]

Summary

• Adds src/permissions/index.test.ts covering summarizeTool(), allow/deny glob matching through checkPermission(), and truncateToolResult() (closes Add unit tests for src/permissions/index.ts #5).

Test plan

• bun run typecheck
• bun test
• bun run build

[github-actions]

🤖 KlaatAI Review Bot (powered by Klaatu, advisory only — a maintainer makes the real call)

Issue match

Fully addresses #5. The diff adds src/permissions/index.test.ts covering summarizeTool for all requested tools (read_file, write_file, edit_file, run_command, apply_patch) and the requested glob-pattern matching. It also exceeds the scope by testing truncateToolResult.

Test coverage

Tests assert exact summary strings and permission outcomes. However, while allowed_commands has both match and non-match cases, denied_commands only tests a match (rm -rf /). A test confirming a non-matching command (e.g., rm file.txt against denied_commands: ["rm -rf /"]) yields "ask" or "allow" is missing.

Correctness concerns

No obvious risks. The run_command truncation test correctly bounds the length and checks for the ellipsis. The tc helper hardcodes id: "call_test", which is fine for these pure logic tests but would require updating if future tests rely on unique IDs.

Verdict

Ready to merge as-is, though adding a non-match assertion for denied_commands would cleanly satisfy the issue's pattern-matching acceptance criteria.

_{This is an automated review to help triage faster, not a gate. Nothing here blocks merging.}

[Prateekgaur]

One gap worth closing before merge: we test that a command matching the deny list gets blocked, but we don't test that a command not on the deny list is left alone. Right now nothing proves the deny rule is precise rather than accidentally blocking more than it should.

Could we add one more case — something like a harmless command (e.g. rm file.txt) checked against a deny list that only contains rm -rf / — and confirm it's not denied? That closes the loop on the pattern-matching behaviour and matches the acceptance criteria in the original issue.