v0.2.7

What's Changed

• fix: sandboxPath silently mishandles literal ~ in file-tool paths by @DrunkkToys in #41

Full Changelog: v0.2.6...v0.2.7

v0.2.6

What's Changed

• fix: stop plan state from bleeding into new chats and going stale aft… by @DrunkkToys in #39
• chore: release 0.2.6 by @DrunkkToys in #40

Full Changelog: v0.2.5...v0.2.6

v0.2.5

What's Changed

• fix: toolsProvider force-bootstrap reset session state every turn by @DrunkkToys in #37
• chore: release 0.2.5 by @DrunkkToys in #38

Full Changelog: v0.2.4...v0.2.5

v0.2.4

What's Changed

• fix: enhance working context — semantic loop guard + distilled fact-memory by @DrunkkToys in #35
• chore: bump version to 0.2.4 by @DrunkkToys in #36

Full Changelog: v0.2.3...v0.2.4

v0.2.3

What's Changed

• fix: context-roll recovery, reasoning-effort no-ops, plan execution, tag sanitization by @DrunkkToys in #34

Full Changelog: v0.2.2...v0.2.3

v0.2.2

What's Changed

• feat: read loaded context length; add reasoning + context settings (v0.2.2) by @DrunkkToys in #32
• docs: add release + add-setting skills, align agent workflow files by @DrunkkToys in #33

Full Changelog: v0.2.1...v0.2.2

v0.2.1

Full Changelog: v0.2.0...v0.2.1

Full Changelog: v0.2.0...v0.2.1

What's Changed

• Fix/vibe bridge round cap by @DrunkkToys in #30

Full Changelog: v0.2.0...v0.2.1

What's Changed

• Fix/vibe bridge round cap by @DrunkkToys in #30
• Fix/vibe bridge round cap by @DrunkkToys in #31

Full Changelog: v0.2.0...v0.2.1

v0.2.0

What's Changed

• feat: active vibe_bridge — model responses via SDK client by @DrunkkToys in #29

Full Changelog: v0.1.1...v0.2.0

v0.1.1

What's Changed

• feat: cascade tests, contract tests, web search fix, pipeline docs by @DrunkkToys in #1
• fix: plugin memory fixes and comprehensive cascade tests by @DrunkkToys in #2
• fix: orchestratorLoop finalText includes tool results by @DrunkkToys in #3
• fix: workspace guard + no hardcoded path by @DrunkkToys in #4
• fix: context window cleanup - memories on disk only by @DrunkkToys in #5
• fix: smart truncation with read_file offset by @DrunkkToys in #6
• fix: stack overflow + 46 cascade tests by @DrunkkToys in #7
• fix: garbage output detection and retry by @DrunkkToys in #8
• fix: force tool_choice required by @DrunkkToys in #9
• fix: auto-save findings + summarize pruned turns by @DrunkkToys in #10
• fix: accumulate all tool results + fix memory saving by @DrunkkToys in #11
• fix: SSD memory architecture — save to disk, keep minimum in RAM by @DrunkkToys in #12
• feat: strategic context injection with system prompt, respond_to_user, loop detection by @DrunkkToys in #13
• fix: remove random context injection — system prompt hack removed by @DrunkkToys in #14
• fix: switch to prediction loop handler with proper context/memory management by @DrunkkToys in #15
• feat: release latest vibeLM state by @DrunkkToys in #19
• feat: restart-safe session state by @DrunkkToys in #20
• Fix restart-safe history dedupe by @DrunkkToys in #21
• Set workspace requires explicit path by @DrunkkToys in #22
• Add explore workspace command by @DrunkkToys in #23
• fix: remove exploration trigger from set_workspace by @DrunkkToys in #24
• fix: remove duplicate tool toggle checkboxes by @DrunkkToys in #25
• fix: memory leak in jsonl cache by @DrunkkToys in #27
• fix: restore tool toggles, remove duplicate tools, rename respond_to_user to amend by @DrunkkToys in #26
• feat: context management optimization - compact earlier, adaptive checkpoint, selective memory by @DrunkkToys in #28

New Contributors

• @DrunkkToys made their first contribution in #1

Full Changelog: https://github.com/DrunkkToys/vibeLM/commits/v0.1.1