v0.1.18

v0.1.18

✨ Features

L3 Replay

• Implement L3 replay primitives — RestoreHook, DriftDetector, schema extensions

General

• Close gaps on 4 partially complete features
• Add per-feature README GIFs, harness improvements, and CI gates

🐛 Bug Fixes

Frontend

• Revert eslint to v9 — react-hooks plugin incompatible with v10

CI

• Add continue-on-error for codecov and deprecated dependency-review action
• Resolve CI lint errors in harness scripts

Tests

• Use UTC dates in analytics DB tests to match repository implementation
• Update test expectation for checkpoint_timestamp in restored session config

Review

• Address all review comments on PR #140
• Address remaining P1/P2 review feedback on restore and event filtering
• Resolve Pyright type warnings in new feature code

Demo

• Re-record demo GIFs with correct feature navigation and slower pacing

📦 Other

Dependencies

• Bump vite from 8.0.3 to 8.0.5
• Bump eslint from 9.39.4 to 10.2.0
• Bump globals from 16.5.0 to 17.4.0
• Bump codecov/codecov-action from 4 to 6

v0.1.17

✨ Features

• Alert lifecycle management, configurable policies, and query performance indexes (#135)
• Add harness improvements — pre-commit hooks, pyright, coverage gate, CI gates, and Claude Code hooks
• Add cost dashboard + trace search demo GIF

🐛 Bug Fixes

• Resolve lint errors and clean up coverage config
• Eliminate N+1 queries in _load_candidate_sessions event_type filter (#132)
• Drop .distinct() from event_type subquery, add tests
• Remove unused sqlalchemy.exists import in storage/search.py
• Address code review findings — critical, major, and minor issues

📦 Other

Documentation

• Split architecture into abstract overview + detailed layer diagram
• Upgrade README architecture diagram with professional styling
• Update README to reference regenerated demo GIFs and screenshots
• Regenerate demo GIFs and screenshots with polished UI

v0.1.16

v0.1.16

Features

CLI

• Add peaky-peek demo command for zero-friction onboarding (seed + launch server + frontend)
• Add peaky-peek seed and peaky-peek serve subcommands

Frontend

• Update WhyButton to use diagnostic API
• Refactor DecisionTree recommendation to useMemo
• Refine replay UX and similar failure search
• Add failure analysis, entity insights, and Hindsight export

SDK

• Elevate to top-tier quality: thread safety, security, API robustness
• Harden SDK with typed schemas and frontend resilience
• Fix tenant isolation and test harness leaks
• Prevent HttpTransport creation without API key

Benchmarks

• Add benchmark seed data enrichment
• Stabilize benchmark test timeouts in CI

Testing

• Restore and fix failure_memory tests
• Stabilize LangChain test suite
• Improve restore flow and seed data validation

Bug Fixes

• Fix session update and collector regressions
• Fix export, drift contract, and request deduping
• Fix contract test import order for CI
• Multiple CI stability fixes (lint, test isolation, timeouts)

v0.1.15

🐛 Bug Fixes

• Fix HTTP transport not being wired up in local-first mode without API key — SDK silently dropped all events instead of sending them to the server
• Disable SDK in benchmark tests to prevent network retries

Full Changelog: v0.1.14...v0.1.15

v0.1.14

What's Changed

Stability & Correctness (Phases 1-2)

• Fix SDK connection leak in shared async client auto-cleanup
• Fix unhandled checkpoint restore errors leaking raw HTTP errors
• Fix SSE disconnect memory leak with CancelledError handling
• Fix duplicate SessionStore interface in frontend
• Add silent failure detection in instrument() auto-patch
• Replace print() with logger in auto-patch transport
• Add proper return types for get_current_context()
• Fix N+1 search pattern loading 500 sessions with all events
• Add rate limiting on API key creation
• Fix buffer silent data drop without caller notification
• Fix fragile replay collapse_threshold default unwrapping
• Fix dangerouslySetInnerHTML XSS risk in ToolInspector
• Remove unused useReplayBreakpoint.ts hook

Quality & Maintainability (Phases 3-4)

• Encapsulate global mutable state in auto-patch transport
• Expose retry config via Config instead of hardcoded values
• Use @abstractmethod in RecordingMixin
• Add DB pool_timeout/pool_recycle for non-SQLite engines
• Add transaction boundaries in trace_routes
• Fix fragile querySelector with stable getElementById
• Improve accessibility (role="meter" replacement, validation cleanup)
• Fix auto-patch side effects at import time
• Improve Config validation flow
• Add CONTRIBUTING.md and update ARCHITECTURE.md

Tests

• Add 32 tests for storage/embedding (tokenize, vectors, similarity)
• Add 25 tests for storage/search (sessions, events, tenant isolation)
• Add tests for trace_routes rollback/commit behavior
• Add tests for engine factory pool settings
• Add tests for comparison_routes and trace_routes
• Add dedicated SessionManager tests

Refactoring

• Remove dead ranking module and unused feature tests
• Deep quality refinement across SDK, backend, and frontend
• Round 2 refinement — dead code removal, component improvements
• Replace falsy-or patterns with explicit None checks
• Improve DX — tooltips, validation, prescriptive hints, keyboard shortcuts

Full Changelog: v0.1.12...v0.1.14

v0.1.13

v0.1.13

✨ Features

Developer Experience

• Add tooltips explaining technical terms (replay_value, drift, policy_shift, ranking factors)
• Add keyboard shortcuts: Ctrl+1/2/3 for tabs, Ctrl+K or / for search focus
• Add prescriptive action hints in FailureClusterPanel mapped to error types
• Add drift direction indicator and remediation guidance in DriftAlertsPanel
• Add cost efficiency hint when per-token cost is notably high
• Add ranking factor breakdown labels in EventDetail
• Add skeleton loading animation to CostPanel
• Add config validation (endpoint URL, max_payload_kb, sample_rate)
• Expand boolean parsing to handle 1/0, yes/no, on/off
• Add actionable HTTP error messages (401/403/404/429)
• Add auto-patch console feedback for CLI visibility
• Enhance error messages with actionable suggestions

Frontend

• Convert LLMViewer and MultiAgentCoordinationPanel empty states to EmptyState component

📦 Other

Refactoring

• Remove dead code and simplify across backend, SDK, and frontend
• Remove dead ranking module and unused feature tests
• Extract severity weights to module-level constant

v0.1.12

v0.1.12

✨ Features

• Add Failure Clustering, Multi-Agent Coordination panels and fix Inspect tab layout
• Improve UI presentation — collapsible sections, clickable checkpoints, tree polish
• Professional UI polish for Inspect tab and DecisionTree
• UX polish for Trace tab, Inspect tab, and cross-tab design system

🐛 Bug Fixes

• Make Inspect tab DecisionTree full-width and auto-height
• Resolve CI failures (imports) — ensure collector submodules accessible, avoid circular imports on Python 3.10

📦 Other

Documentation

• Replace feature screenshots with per-feature GIFs in README
• Re-record feature GIFs with updated UI

Refactoring

• Deslop codebase — remove dead code, consolidate duplicates, fix DDD boundaries, polish UI layout
• Deduplicate _event_value across collector modules

Performance

• Optimize performance across backend, frontend, and SDK

Testing

• Add 94 tests for untested high-risk modules

v0.1.11

v0.1.11

✨ Features

• Implement medium-term features — adaptive replay, exploration, conversation, live dashboard
• Implement 5 research-backed quick wins

🐛 Bug Fixes

• Auto-increment session.errors when error events are added
• Use atomic SQL increment for session.errors in add_event
• Polish 6 issues for HN launch — drift threshold, seed data, benchmark calls
• Resolve CI failure (version) — sync expected version to 0.1.10
• Resolve CI failure (version) — sync hardcoded version in SDK init.py
• Resolve CI failure (lint) — break long line in seed_data.py
• Resolve CI failure (lint)

📦 Other

Dependencies

• Bump typescript from 5.9.3 to 6.0.2 in /frontend
• Bump vitest from 4.1.1 to 4.1.2 in /frontend

Documentation

• Add polished UI screenshots for README
• Add updated interactive course for Peaky Peek
• Add GitHub icon/link to course header
• Update course description with key features
• Update README screenshots and title
• Add scientific papers section and fix safety screenshot
• Remove Safety Audit Trail section
• Add research-backed action plan and paper analysis

Refactoring

• Deslop — trim verbose docstrings and remove duplicated formatTimestamp
• Derive version from installed package metadata
• Add 174 collector tests, extract UI components, fix test regressions

Testing

• Enhance auto-patch adapter tests — verify event data fields and missing paths

Style

• Polish UI for more professional appearance

v0.1.10

v0.1.10

✨ Features

UI

• Distill UI — restructure tabs, unify color tokens, add panel hierarchy

🐛 Bug Fixes

Seed data

• Enrich seed data — add tokens, costs, retention, alerts, fix error semantics
• Update dev proxy port and add missing db commits in seed script

CI

• Update benchmark tests for explicit error events
• Auto-fix ruff import ordering and unused imports

Config

• Make dev proxy port configurable via API_PORT env var

📦 Other

Tests

• Add regression tests for 6 discovered issues

Chores

• Bump version to 0.1.9
• Bump version to 0.1.10

v0.1.9

v0.1.9

✨ Features

• Pre-launch polish — simplified API, Docker, README, adapter improvements

🐛 Bug Fixes

• Resolve open issues #107-#113 — simplify complexity, add logging, extract imports
• Resolve drift alert bugs and add comprehensive test coverage
• Use parameterized query in _repair_legacy_sqlite_schema
• Suppress bandit B608 false positive in analytics_db

📦 Other

Refactoring

• Deslop SDK, server, and test suite — remove ~450 lines of duplication and dead code
• Decompose intelligence module and expand unit test coverage
• Complete test split — remove monolith, fix lint, update doc refs
• Deslop frontend types, emitter, and config — deduplicate shared utilities

Chores

• Bump SDK version and test to 0.1.8
• Untrack .omc/state from version control