v0.12.0

aislop v0.12.0 adds a terminal-native agent workflow for scanning, repairing, and reviewing AI-generated code without leaving your local checkout.

Highlights

• Added aislop agent for provider-aware repair sessions from the terminal.
• Added agent session, monitor, apply, watch, stop, and provider commands.
• Added live session state with edited files, score targets, token usage, and review steps.
• Improved terminal output, command suggestions, scoped scans, and rule impact metadata.
• Calibrated scoring so low-impact style signals are softer while real findings stay visible.
• Improved Python and Go dependency/import detection to reduce noisy false positives.

Fixes

• Fixed agent worktree diffs so staged and untracked files are included correctly.
• Fixed monitor debounce handling so stable changes trigger the settled scan/repair cycle.
• Fixed telemetry coverage, subdirectory scope handling, monitor stop behavior, and installer stability.
• Fixed zero-debounce monitor mode so --debounce 0 reacts immediately to the first changed snapshot.

Validation

• 144 test files passing
• 1,276 tests passing
• Self-scan score: 100/100
• 0 diagnostics
• npm pack --dry-run verified aislop@0.12.0 with 26 files