v0.7.9

What's changed

• feat: support per-entry variants in auditor fallback models

Auditor fallback models can now be specified as { model, variant } objects so each fallback carries its own reasoning/thinking variant; string entries still work and run with no variant (the primary auditorVariant is never inherited by fallbacks).

v0.7.8

Breaking changes

• The sandbox runtime now uses the sbx CLI instead of Docker. Sandbox execution and the container template have migrated from Docker to sbx create/sbx exec. Custom mounts now require absolute host paths, and legacy Docker sandbox configuration emits warnings.

What's changed

• feat: migrate sandbox driver from Docker to the sbx CLI (#82)
• feat: fall back to alternate auditor model on provider limit (#81)
• fix(sandbox): canonicalize paths when detecting mount overlap (#83)

v0.7.7

• feat: validate stored plan structure and gate autonomous launches on readiness (#80)

v0.7.6

• feat(dashboard): unify dashboard bind host/port config resolution (#78)
• chore: bump version to 0.7.6

v0.7.5

• feat(dashboard): add multi-level repo shell with searchable loop navigation (#75)
• chore: bump version to 0.7.5

v0.7.1

feat: allow plan-adjust to edit the section under audit in place

• Plan adjustments: new optional currentSection argument lets the auditor revise the section currently being audited in place, preserving its progress (status/attempts). The existing sections argument becomes optional; omit it to leave future sections unchanged.
• Service validation: rejects calls with neither sections nor currentSection, refuses to edit an already-completed section, and snapshots from the earliest changed index for accurate before/after amendment records.
• Tool, prompt, and docs: updated plan-adjust tool description, section audit prompt, auditor-loop addendum, and docs (loop-system, modules, tools) to reflect the new in-place edit capability.
• Tests: covers validation errors, in-place editing preserving progress, omitting sections, combined current+suffix replace, and empty-suffix removal.

chore: bump version to 0.7.1

v0.7.0

feat: add loop FSM transitions, plan amendments, and audit rewind (#72)

• Loop FSM transitions: state-machine transition definitions with persistence via new loop_transitions repository.
• Plan amendments: new plan_amendments repository tracks auditor edits to remaining plan sections; adds plan-adjust tool.
• Section plans repository: dedicated persistence for per-section plan state.
• Audit rewind: loop-final-audit rewind support for re-running audits.
• Loop runtime/service: integrated transition logging and state management across runtime, service, and hooks.
• Dashboard: new machine-graph component for loop state visualization, plus data/render updates.
• Migrations 141-143: extend loop phase-check, create loop_transitions and plan_amendments tables.
• Comprehensive tests: runtime transition logging, section advancement, transitions/amendments repos, execution restart, loop service, storage migrations.

fix: resolve execute goals from session context

v0.6.2

feat: add provider limit abort handling for loop runtime

• Provider limit detection: new module detects rate-limit and abort signals during prompt execution, enabling graceful loop termination.
• Abort signal propagation: integrated provider limit checks into the runtime prompt flow with watchdog monitoring.
• Fallback permission patching: subagent sessions in loop worktrees now receive the loop ruleset via a fallback path when session.created events are filtered by instance directory.

feat: add fallback loop permission patching for filtered session.created events

• Refactored single-event hook into LoopPermissionPatcher with ensurePatched fallback called from chat.message and tool.execute.before hooks.

v0.6.0

fix: harden remote loop launches against stale branches, clock skew, and silent errors

• Pinned-branch validation: a SHA-pinned launch now fails loudly when a same-named branch already exists at a different tip, instead of silently running old code.
• Clock-skew normalization: configure re-stamps workspaceCreatedAt and forgeLoop.pendingAttachStartedAt with the server's clock so the attach grace window is not expired by skew between the launching machine and the attach server.
• Surface loop launch errors: launchTuiLoop returns { error } with the underlying cause instead of null; the remote launcher cleans up the sync ref on failure.
• Terminal-name conflict attach: a fresh TUI launch colliding with a terminal loop row no longer silently skips; it routes through attachLoopToSession so the conflict surfaces as a toast + workspace removal.

v0.5.7

simplify goal-loop executor binding; add sandbox container placeholder