v0.14.0

The biggest change in this release is the initial version of the REPL and this is the first release where I didn’t have to do a single handoff - it was completely handled by the workflow

What's changed

• 🔖 Prepare v0.14.0 release (#1097) @taras
• Continue initial planning through original role discovery (#1092) @taras
• 🐛 Separate passive-supervision bootstrap from subscription checks (#1096) @taras
• 🐛 Prevent passive supervision cleanup from hiding test failures (#1094) @taras
• 🐛 Keep foreground command output inside its REPL entry (#1093) @taras
• Require joined review evidence for workflow delivery readiness (#1064) @taras
• Supervise workflow phases through the existing follow stream (#1058) @taras
• Refuse conflicting Claude identity during startup (#1090) @taras
• Require opt-in for absolute host directories and preserve approved imports (#1074) @taras
• Support standard ACP agents with explicit session guarantees (#1085) @taras
• Add reusable local ACP agent qualification (#1079) @taras
• Document host-catalog selection for Fable reviews (#1083) @taras
• Use host Claude Code throughout the retained ACP integration (#1082) @taras
• Read saved command diagnostics through Journal queries (#1078) @taras
• Import approved host assets into the REPL workspace (#1071) @taras
• Add direct live-child REPL submission with streamed progress (#1070) @taras
• Acquire REPL repositories inside the session workspace (#1068) @taras
• Enable existing assertions in ordinary REPL entries (#1067) @taras
• Make REPL syntax discovery reflect entry availability (#1065) @taras
• Support nested Output declarations through scoped rendering (#1061) @taras
• Preserve multiline pastes as one Sidekick message (#1063) @taras
• Keep Sidekick typing responsive as history grows (#1060) @taras
• Recover Sidekick without replaying interrupted work (#1057) @taras
• Preserve literal bindings inside passive code fences (#1055) @taras
• Add a zero-correction option to workflow phases (#1056) @taras
• Run parent-owned headless XMD children with live attachment (#1044) @taras
• Focus independent review on correctness, minimalism and performance (#1049) @taras
• Wait for completed Draft typing in the presentation journey (#1047) @taras
• Adopt verified adapter publication after a recovery race (#1045) @taras
• Keep History controls reachable as entries accumulate (#1041) @taras
• Refresh measured runner weights after ACPX integration (#1036) @taras
• Adopt ACPX 0.21.1 with strict conversation resume (#1028) @taras
• Continue Sidekick from execution results with Follow (#1034) @taras
• 🤖 Route implementation work from prepared scope (#1033) @taras
• Queue Sidekick messages while it is busy (#1030) @taras
• Scroll REPL readings with the mouse wheel (#1027) @taras
• Wait for completed history in the Sidekick migration test (#1024) @taras
• Navigate REPL regions and edit text with visible carets (#1022) @taras
• Test REPL behavior through terminal page objects (#1016) @taras
• Avoid redundant workflow corrections and manual phase setup (#1020) @taras
• Read narrated workflow results without repeating completed work (#1014) @taras
• Validate Planner APIs before Implementor migrations (#1012) @taras
• Measure test weights in separate jobs and retain partial results (#1011) @taras
• Simplify REPL tests with a shared Termless screen (#1010) @taras
• Give the REPL a responsive workspace and independent Stops (#1004) @taras
• Let trusted hosts stop executions and join owned cleanup (#1008) @taras
• Reduce memory use when opening and continuing REPL histories (#997) @taras
• Recover ownership of a stuck agent session (#996) @taras
• Connect ACP clients to a live REPL Sidekick (#989) @taras
• Give sibling Session catalog isolation checks a 60-second timeout (#995) @taras
• Support model and effort configuration in TestAgent scenarios (#988) @taras
• Mount external directories in the REPL workspace (#991) @taras
• Keep long REPL output readable (#990) @taras
• Run component tests from the Sidekick (#986) @taras
• Load TypeScript components from their selected source (#983) @taras
• Use the REPL session workspace for Sidekick entries (#979) @taras
• 🧪 Wait for REPL questions before asserting lifecycle (#981) @taras
• 🧪 Make the REPL permission test wait for its running sibling (#980) @taras
• 🐛 Keep REPL source ownership across sequential nested calls (#977) @taras
• Open pull requests and resume feedback reads in xmd repl (#970) @taras
• Use runner measurements for CI test allocation (#973) @taras
• ✨ Load external components with xmd repl --include (#972) @taras
• 🐛 Keep xmd repl usable after repeated component calls (#969) @taras
• 🚨 Retain complete Workflow pull-request reads after cleanup (#966) @taras
• Prevent automatic maintenance from escaping generated Git fixtures (#965) @taras
• ✨ Select structured data with Query and render YAML (#934) (#962) @taras
• Make the <Git.Push /> authentication cancellation test reliable (#950) (#959) @taras
• Define the security model and hook API contract (#954) @taras
• Inspect saved REPL entry evidence from the Sidekick (#932) (#955) @taras
• ♻️ Share one <Git.Push /> and <Issue> between the REPL and every other command (#951) @taras
• Require necessity checks in agent plans and reviews (#953) @taras
• ✨ Show a window of rendered text with (#942) (#952) @taras
• ✨ Read tracked files, changes, diffs and committed text (#940) (#944) @taras
• ✨ Reuse root JSON across REPL entries and cold reopen (#939) (#943) @taras
• Add resumable <Git.Push /> to REPL entries (#937) @taras
• Record accumulated architecture and agent-writing guidance (#936) @taras
• ✨ Keep a REPL execution's chat beside its entries (#883, PR 2 of 2) (#921) @taras
• ✨ Evaluate, preview and converse as a trusted host (#883, PR 1 of 2) (#890) @taras
• ✨ Stage, switch and commit in REPL entries (#927) @taras
• File and read GitHub issues in resumable REPL entries (#926) @taras
• ✨ Resume REPL entries in managed repositories and worktrees (#925) @taras
• Resume REPL entries without repeating completed file operations (#902) @taras
• Replace host evaluation profiles with Component.generated (#181) (#920) @taras
• 💥 Let a component say what its own spellings are offered as (#181) (#919) @taras
• 🧪 Identify core refusals and retained test observations (#916) @taras
• test: isolate npm packaging output per invocation (#918) @taras
• test: observe Web response settlement and preserve failure diagnostics (#917) @taras
• 🧪 Make lifecycle and publication tests observe complete outcomes (#915) @taras
• 📝 Clarify execution contracts and repair compiled PR review (#910) @taras
• ✨ One place for a run's setup, and a Syntax description anything may compose (#912) @taras
• 📝 Make the architecture documents easier to read and navigate (#913) @taras
• Simplify project comments and enforce banned words (#908) @taras
• 🔒 One document filesystem, and one policy on it (#181) (#906) @taras
• 🔒 Canonical core owns File, File.Delete and Dir (#181) (#905) @taras
• 🔥 Remove unused repository code and reject new dead code in lint (#903) (#904) @taras
• Revert "Add repository operations and recorded-result replay to xmd repl" (#900) @taras
• Add repository operations and recorded-result replay to xmd repl (#899) @taras
• ⚡️ Reuse the layout work REPL typing and focus did not change (#881) (#889) @taras
• Navigate recorded History positions with a measured REPL rail (#881, 3/3) (#887) @taras
• Show live component lifecycle and generated source in the REPL (#881, 2/3) (#886) @taras
• Match REPL colours, pane surfaces and drawers to the reference (#881, 1/3) (#885) @taras
• 🔨 A bench for the REPL's terminal, and what it found (#884) @taras
• 💥 Give a workflow run one definition, and no version at all (#880) @taras
• 📚 Define REPL history in architecture terminology (#879) @taras
• 💄 Draw the REPL as the things it is showing (#878) @taras
• 🐛 Keep the catalog reveal the restoration claim moved (#877) @taras
• ✨ Measure the REPL's terminal layout before it admits a frame (#876) @taras
• ✨ Make xmd repl operable and self-explaining from its terminal (#870) (#871) @taras
• ✨ Slices C and D of #827: a draft, an Entries catalog, and the journey that proves them (#869) @taras
• ✨ Slice B of #827: own one REPL entry task at a time, with entry-aware Agent turns (#867) @taras
• ⚡ Measure test weights at b58d937 (#868) @taras
• ✨ Slice A of #827: project sequential REPL entries from one Journal (#866) @taras
• ⚡ Measure test weights at f7a1063 (#865) @taras
• ✨ Slice E of #854: run the packaged Plan in xmd repl, under one immutable profile (#864) @taras
• ✅ Wait for every Prompt before counting the whole chronology (#863) @taras
• ✨ Slice D of #854: present Agent conversations, turns and permissions in the REPL (#861) @taras
• ✨ Slice C of #854: ask bounded Elicit forms in the REPL (#860) @taras
• ✨ Slice B of #854: run live Agent turns and permission waits in the REPL (#859) @taras
• ✨ Run spawned document work concurrently with (#855) (#856) @taras
• ✨ Slice A of #854: retain a safe Agent permission audit and project Agent conversations (#857) @taras
• ✨ Slice E of #848: run and reconstruct one XMD entry in a REPL (#853) @taras
• ✨ Slice D of #848: draw and own a REPL terminal (#852) @taras
• ✨ Slice C of #848: reconcile the REPL component tree (#851) @taras
• ✨ Slice B of #848: run and resume one REPL execution (#850) @taras
• ✨ Slice A of #848: project REPL views from durable history (#849) @taras
• 🐛 Build npm packages from their local release closure (#847) @taras