Policy Scout added a read-only Cleanup Dry-Run card to the experimental Tauri UI, previewing cleanup targets without deleting files.
Policy Scout added the Tauri Cargo.lock file so the experimental desktop app has reproducible Rust dependency resolution.
Policy Scout refactored the experimental Tauri Overview into small read-only React components while preserving existing cards and adapter boundaries.
Policy Scout added a read-only Tauri Audit Stats card backed by an allowlisted get_audit_stats command and audit stats JSON output.
Policy Scout added a read-only Tauri Reports List card backed by an allowlisted list_reports command and report list JSON output.
Policy Scout cleaned the experimental Tauri adapter by renaming the shared CLI JSON response type and centralizing allowlisted CLI execution.
Policy Scout added a second read-only Tauri card showing local data status through an allowlisted get_data_status backend command.
Policy Scout added its first experimental Tauri scaffold with one read-only Doctor status card backed by an allowlisted get_doctor_status command.
Policy Scout added a Tauri adapter boundary spec defining allowed UI backend commands, forbidden capabilities, CLI allowlists, validation rules, and first scaffold sequence.
Policy Scout added decision and risk details to blocked run --json responses while preserving execution blocking, exit codes, and JSON compatibility.
Policy Scout added a backward-compatible duration_ms alias to eval JSON output while preserving execution_time_ms for existing consumers.
Policy Scout inventoried current CLI JSON output shapes and added current-contract coverage so future JSON API v1 changes can be made from tested evidence.
Policy Scout replaced deprecated UTC datetime usage in approval tests with timezone-aware Python 3.12-safe calls while preserving behavior and JSON contracts.
Policy Scout clarified local CLI invocation patterns and documented why PYTHONPATH is expected for subprocess test isolation, with no behavior or JSON changes.
Policy Scout added JSON API v1 planning for future stable CLI, Tauri, and local adapter contracts without changing runtime behavior or current JSON outputs.
Policy Scout added a Tauri component/data contract map linking future UI screens to CLI JSON commands, key fields, IDs, redaction states, and deferred mutation boundaries.
Policy Scout added compact Tauri screen wireframes for nine read-only UI screens with CLI JSON mappings, redaction rules, and deferred mutation boundaries.
Policy Scout added a compact Tauri UI Plan v0 defining a read-only dashboard path, CLI subprocess boundary, allowlisted wrapper model, and deferred mutation screens.
Policy Scout added JSON contract tests for run, audit show/request/type, and eval outputs, strengthening CLI machine contracts for future Tauri readiness.
Policy Scout documented preview-only data cleanup planning in README and status docs, including supported temporary targets and the no-deletion safety boundary.
Policy Scout polished report list human output: empty Created fields now show unknown, report lists include a report show hint, and JSON output remains unchanged.
Policy Scout stabilized approval execution tests by replacing expired hardcoded fixture timestamps with dynamic future expirations.
Policy Scout added --type filtering to report list, making large local report collections easier to browse by command, sandbox, project sweep, or quick sweep report type.
Policy Scout aligned IMPLEMENTATION_STATUS.md with current alpha state: 591 tests, 44 evals, data command, npm-only sandbox execution, current migration syntax, and updated limitations.
Policy Scout moved compact source docs into docs/compressed to match AGENTS.md reading order and added the scenarios/evals decision source.
Policy Scout added a README Mermaid safety-boundary diagram and consolidated visual/diagram guidance into a compact source doc without new render dependencies.
Sweep JSON reports now explicitly mark redaction_applied in metadata, aligning sweep output with sandbox/report JSON contract expectations.
Policy Scout added a read-only data command showing local paths, existence, and counts for audit, reports, sandbox, demo, migration, and backup state.
Policy Scout clarified sandbox install vs migration help, prints report IDs after sandbox review, and adds redaction/file-change metadata to sandbox JSON reports.
Policy Scout now declares pytest as a dev extra and CI installs .[dev], fixing the clean GitHub Actions failure while keeping runtime dependencies minimal.
Policy Scout README now includes alpha quickstart, doctor/demo flow, safety-gate examples, JSON mode, current test/eval status, and honest local-first limitations.
Policy Scout added JSON contract tests for doctor/check/audit/report/sweep and fixed check --json command redaction so secret-like command values are replaced with canonical placeholders.
Policy Scout added a safe local demo command that creates a fixture workspace, runs allow/sandbox/deny/credential/destructive checks, and sweeps the fixture without real execution, package installs, network access, or secret reads.
Policy Scout’s commit and bump gate now documents the required Project field so blog.bumper routes future PASS COMPLETE messages to the policy-scout project instead of falling back to the global default.
Root AGENTS.md was replaced with a Policy Scout-specific persistent agent reference. The new operating file updates the reading order to the compiled implementation docs, preserves repo-specific command/test/isolation details, and reflects the current CLI-first alpha state.
Policy Scout now has minimal GitHub validation and a documented commit/blog coordination gate. CI checks doctor, evals, and the full test suite on push and PR to main; PRs now include safety-focused review prompts; the bump gate documents the reviewed code-plus-dev-log workflow.