Changelog

A complete history of features, fixes, and changes in H·AI·K·U (formerly AI-DLC).

RSS
v10.3.0
Added
  • Optional stages (design, operations, security) can be dropped at elaboration start if they're not needed for the intent.
  • Stage briefs (BRIEF.md) document the plan at stage start and the results at completion.
  • Provider authentication: agents can authenticate with GitHub or GitLab, with credentials stored locally for PR operations and CI checks.
  • Feedback can be moved between stages when findings belong elsewhere.
  • Runtime verification proof uploads directly to delivery PRs, showing test evidence to reviewers.
v10.2.1

No notable changes.

v10.2.0
Added
  • Runtime verifier now captures video and screenshots of verification runs, automatically uploading proof to the stage PR for inspection and review.
Changed
  • Discrete-hybrid mode now creates per-stage PRs only for stages with external review gates, keeping continuous stages on the intent-main PR to reduce PR clutter.
v10.1.0
Added
  • Intents now hold at pending seal until merged into the default branch, showing delivery status at a glance
  • Browse displays unit creation dates and activity timelines; hat/fix histories are now interactive steppers; outputs and agent definitions open inline
Changed
  • Browse list views now use optimized queries to handle large intent counts
Fixed
  • Browse no longer drops intents past index 100
  • Unit metrics and code inputs/outputs now display accurate data from real sources
  • Engine state is protected during branch-to-stage merges
v10.0.0
Added
  • Optional stages: design, operations, and security are now droppable per intent; libdev studio deprecated for optional release stage.
  • Feedback severity-gated fix loops: only blocking findings trigger fixes; non-actionable findings handled separately.
  • Review agents, hats, and fix hats now resolve via project/studio/stage hierarchy, enabling per-studio or per-stage customization without forking.
  • Discovery artifacts now persist across intents and auto-load as starting points for new work.
  • Stage observations auto-record and summarize at intent close with studio-defined perspectives; opt out via reflection: false.
Changed
  • Unit and fix-chain isolation: code runs on ephemeral git branches per unit/fix, pushed on advance, merged at terminal with recovery on pickup.
  • Delivery verification gates on CI meaningfulness (not just green) and merge readiness; escalates to human when provider CLI access is missing.
v8.0.0
Added
  • /haiku:debug skill to inspect, diagnose, and recover intents stuck in error states
  • Iteration bounds on haiku_run_next to prevent infinite loops
Changed
  • Intent state format (automatic v7 → v8 migration)
v7.0.3
Fixed
  • Workflow engine detects infinite loops and halts gracefully, preventing deadlock on long or complex intent runs.
v7.0.2
Fixed
  • Feedback now correctly persists targeting metadata on creation, enabling close-hook workflows to properly clear approvals when resolving findings.
v7.0.1
Fixed
  • Plugin inlining process now handles multi-line template calls, restoring functionality from 7.0.0.
v7.0.0
Added
  • Automatic migration from v5 to v6 workflows, preserving all intent history and configuration during upgrades.
Changed
  • Improved formatting and clarity of review feedback comments in bot-generated responses.
  • Enhanced intent completion review interface for clearer workflow status visualization.
Fixed
  • Session API integration routes now support broader external tool connectivity scenarios.
  • GitHub and GitLab provider integrations corrected for platform-specific requirements.
v6.0.0
Added
  • Intent-scope quality gates validate completion criteria across entire intents.
  • Discovery phase can now iterate through multiple questioning rounds to refine investigation scope.
Changed
  • Revisit workflows now integrate directly into the primary flow; /haiku:revisit skill removed.
Fixed
  • Drift detection no longer falsely blocks progress when open feedback exists.
  • Feedback moving across stages now preserves attachments and maintains numbering.
  • Execute phase no longer permits empty iteration counts on units.
v5.0.3
Added
  • Workflows can now resume from intermediate stages and continue work from where they left off.
  • Unit outputs auto-populate from prior iteration bolts when available, reducing manual data entry.
Fixed
  • Engine now self-heals from workflow drift conditions without requiring manual intervention.
  • Feedback advancement through review hat sequences is now properly tracked and dispatched.
Changed
  • Review UI refactored with expanded feedback panels for clearer visualization and improved mobile responsiveness.
v5.0.2
Fixed
  • Fixed 404 errors when switching tabs in the review UI and resolved spec-gate routing issues that blocked stage transitions.
v5.0.1
Fixed
  • Fixed schema migration for users upgrading plugin from v4.x to v5.0.1.
v5.0.0
Added
  • New haiku_await_gate tool for blocking intent progression while waiting for external reviews or customer response
Changed
  • /haiku:reset skill split into /haiku:reset-stage (for single stages) and /haiku:reset-intent (for full intents)
  • Review UI now visually distinguishes user-authored feedback from agent-authored findings
  • Stage transitions now invalidate downstream approvals when revisiting earlier stages, preventing stale blocks
Fixed
  • Stage reset now properly clears all gate markers and approval state
  • Improved cursor position tracking during hat progression and v3-to-v4 intent migrations
v4.5.6
Fixed
  • Review stepper now scopes correctly and derives stage phase from disk for accurate progress tracking.
  • Removed "(seen)" indicators from the review interface for a cleaner display.
v4.5.5
Fixed
  • Revisiting an earlier stage now invalidates downstream approvals, preventing stale approvals from blocking workflow progression.
v4.5.4
Fixed
  • User-gate decisions now process inline during workflow advancement, eliminating a separate confirmation step.
  • The workflow exits cleanly when the browser disconnects instead of leaving the system in a hanging state.
v4.5.3
Fixed
  • Agent-authored feedback now closes automatically instead of remaining open in the workflow
  • Review gates only block progression on human-authored feedback, not on agent-generated findings, reducing unnecessary holds
  • Worktree switching now works freely even when previous worktrees are locked, improving the handoff experience between branches
v4.5.2
Fixed
  • Stage approval is now blocked when open feedback remains, ensuring all findings are addressed before advancement.
  • Feedback sidebar in the review interface is more reliable when managing multiple findings across stages.
  • Stage cursor position now correctly tracks filesystem-based approvals across branch boundaries.
  • Decision mode type safety improved by defaulting ad-hoc mode to false.
Changed
  • Intent and unit detail views on the website display richer stage progression metadata.
v4.5.1
Fixed
  • Workflow no longer gets stuck in merge loops when a stage branch is an ancestor of main after migration.
v4.5.0
Added
  • haiku_stage_reset tool to reset and restart a stage from its initial state.
  • Loop-guard diagnostic capture for troubleshooting workflow blockages and stuck iterations.
Fixed
  • Post-migration merge-debt loops automatically resolved, eliminating no-op merge cycles after v0→v4 upgrades.
  • Stage review UI merge-state visualization consistency.
Changed
  • haiku_run_next now handles mid-merge blocking and parallel gate resolution more robustly.
v4.4.1
Fixed
  • Fixed workflow stalls that occurred after intent migration when merging branches with identical content.
v4.4.0
Added
  • Workflow auto-repairs missing review and approval stamps on pre-tick with deadlock detection to prevent recovery loops.
  • Intent creation is now available as a standalone tool with metadata-pollution safeguards.
  • Automatic migration from v0 to v4 workflow state.
v4.3.2
Fixed
  • Cursor tracking is now purely disk-based, removing unnecessary git context switches during workflow execution.
v4.3.1
Fixed
  • Merge-stage workflows no longer loop unexpectedly when all approval gates have been satisfied — the all-approvals-stamped flag now correctly takes precedence over iteration shape.
v4.3.0
Added
  • New haiku_discovery_complete tool for agents to mark discovery work as done during elaboration.
Fixed
  • Discovery items now correctly scope to the active stage, preventing discovery work from earlier stages from being incorrectly reprocessed during later stage transitions.
v4.2.3
Fixed
  • Resolved three critical workflow engine bugs that blocked fresh intent creation and prevented progression in active workflows, improving worktree handling and loop detection.
v4.2.2
Fixed
  • Cursor position tracking no longer relies on git history searches, making intent state recovery faster and more reliable in interrupted or partially-corrupted workflow scenarios.
v4.2.1
Fixed
  • Plugin now automatically recovers corrupted workflow state by inferring stage merge history from git, preventing stuck workflows.
v4.2.0
Added
  • Stages can now independently gate elaboration, verifying design work is ready before execution
  • haiku_stage_elaboration_seal and haiku_intent_seal tools to explicitly mark elaboration and intent completion
Changed
  • Elaboration now gates per-stage instead of as a unified design phase
Fixed
  • Eliminated infinite loops in dispatch, approval, and intent-review workflows
  • Discovery now correctly identifies artifacts without relying on metadata
v4.1.4
Fixed
  • Engine no longer gets stuck when handling review dispatch, approval, and intent-review transitions; workflows now progress reliably through quality gates.
v4.1.3
Fixed
  • Discovery artifacts are now tracked by their actual existence rather than frontmatter metadata stamps, eliminating edge cases where the engine lost visibility into completed discovery work and failing to advance the workflow.
v4.1.2
Fixed
  • Intent migration from v0 to v4 now completes reliably, unblocking projects that started with the initial schema.
  • Workflow engine correctly handles merge operations when stage branches are created late in the lifecycle.
  • Review gate schema validation works across all intent schema versions without false rejections.
  • Workflow cursor properly tracks stage position through merge operations, preventing state regressions on subsequent ticks.
v4.1.1
Fixed
  • Migration from v0 to v4 now displays a completion banner and correctly resolves schema-target paths in the new file structure.
  • Blog post routes no longer encounter conditional dynamic export errors.
v4.1.0
Added
  • Create draft pull requests from intents with automatic stage-branch management for improved multi-developer collaboration
v4.0.2
Fixed
  • Changelog entries now correctly handle multi-line bullets, inline markdown formatting, and version bumps with no user-facing changes.
v4.0.1
Fixed
  • Bump-type detector now lets pass 2 (Claude classification) escalate minor → major when the diff is clearly a major rewrite, not just patch → *. Caught the v3.17.0-was-supposed-to-be-v4.0.0 misclassification at the source so the next cursor-class refactor that arrives with a feat: prefix gets caught instead of slipping through.
v4.0.0
Breaking changes
  • Plugin major bump. The on-disk schema generation is now v4. The migrator at packages/haiku/src/orchestrator/migrations/v0-to-v4.ts runs on first read of any pre-v4 intent: strips deprecated FM, normalizes past-tense iteration results ("rejected" → "reject", "advanced" → "advance"), synthesizes user approvals on completed units, deletes per-stage state.json files, and drops pre-v4 baseline noise. Idempotent on re-run.
  • Per-stage `state.json` removed. The cursor reads frontmatter directly. Any tool, hook, or external integration that read or wrote stages/*/state.json must switch to FM-driven derivation.
  • `feedback_id` is an integer (1..999) at the wire boundary. Every haiku_feedback_* MCP tool input takes a number; the canonical display form FB-NNN only appears in human-facing prose and filenames. Tool calls passing feedback_id: "FB-001" (string) are rejected at the AJV gate with <tool>_input_invalid.
  • Numeric prefix padding for files. New unit and feedback files pad to 3 digits (unit-NNN-slug.md, NNN-slug.md). Existing 2-digit names still resolve via numeric-prefix matching, so in-flight intents migrate transparently.
  • Auto-resolve refusal in git mode. haiku_run_next no longer falls through to "the only active intent on disk" when the current branch isn't an intent branch. Either pass intent explicitly or git switch haiku/<slug>/main. Filesystem-mode intents keep the fallback (no branch signal).
  • `upstream_stage:` field and pre-tick triage gate are gone. Cross-stage feedback routing is by file location now; classifier hat (first entry in fix_hats:) calls haiku_feedback_set_targets on dispatch.
  • `manual_change_assessment` action removed. Drift events become drift_detected directly via Track C of the cursor walk.
Added
  • The cursor. packages/haiku/src/orchestrator/workflow/cursor.ts with derivePosition(slug) walking Track C → B → A every tick. Pure observation; same disk state, same answer, every time.
  • `withIntentMainLock` serialises stage→intent-main merges. Wraps mergeStageBranchIntoMain at all three call-sites (haiku_run_next, side-effects.ts pre-stage cleanup, side-effects.ts finalize-stage).
  • TypeBox + AJV input gates on every MCP tool with stable named error codes (<tool>_input_invalid). One source of truth per shape.
  • Width-flexible filename lookup in unitPath() and findFeedbackFile(). Legacy 2-digit and new 3-digit names both resolve.
  • Studio-level review agents + fix hats (plugin/studios/{studio}/ review-agents/ and plugin/studios/{studio}/fix-hats/) for intent-completion review after the final stage gate. Opt out with intent_completion_review: false.
  • Spec review hard gate between every stage's execute phase and the parallel quality review. Engine-owned, no per-studio mandate file, no opt-out.
  • `clarifications` field on intent FM for per-stage clarify gates.
  • `fix_hats:` mechanism on STAGE.md — dispatch a hat sequence per-finding against open feedback. Terminal feedback-assessor hat independently decides closure.
  • Drift detection rewrite. Body-sha256 for unit specs, content- aware sha for outputs (markdown body-hashed; binaries full-file). Repo-relative output paths join against repoRoot not intentDir (was a silent gap on real code artifacts).
Changed
  • Phases are cursor-derived, not stored. phase: is no longer a field on stage state; the cursor reads on-disk shape (units present? hats progressed? reviews signed?) and decides the next action.
  • Stage loop reframed as five phases: elaborate, execute, review, approve, merge. Pre-execution adversarial review removed; review fires after execute.
  • Five operating modes: continuous, discrete, discrete-hybrid, autopilot, quick. The HITL/OHOTL/AHOTL taxonomy is deprecated; user involvement is shaped by the per-stage gate: field plus the invocation skill.
  • Cross-stage FB rewalk via Track B walks every prior stage before the current one. An upstream finding gets attention before forward progress resumes.
Fixed
  • All issues from PR #323's nine review rounds (22 total). Notable: withIntentMainLock not called, result: "rejected" not normalized in migrator, close_feedback handler used wrong filename prefix, parseFbIdFromFilename raw-basename fallback caused infinite loops, exact semver compare in run-tick.ts would have broken every v4 intent on the first patch bump.
v3.16.2
Fixed
  • Gate status badges no longer appear in ad-hoc review panes, reducing visual clutter in temporary review contexts.
v3.16.1
Fixed
  • Visual answer feature now properly supports await mode and handles cancellation without losing state.
v3.15.1
Fixed
  • Workflow state drift detection is now more accurate and won't incorrectly flag valid workflow states as corrupted.
v3.14.2
Fixed
  • The await gate now correctly detects when an external event has occurred, unblocking workflows that were waiting for it.
v3.14.1
Fixed
  • Gate review now submits to external review systems without blocking workflow progression, and correctly identifies intent-level versus stage-level reviews.
v3.14.0
Added
  • Choose operating mode (discrete, continuous, or hybrid) when creating an intent, with constraints enforced based on the selected studio.
  • Select which stage to start in before advancing from elaboration, enabling flexible entry points into staged workflows.
  • New /haiku:change-mode skill to switch between discrete and continuous modes during an active intent.
  • Automatic recovery from dirty tree conflicts when the workflow encounters uncommitted local changes, reducing manual intervention.
Changed
  • Intent creation now includes a validation gate that reviews the intent specification before advancing to the first stage.
  • Elaborate phase prompts now inject structured discovery and output template guidance directly into the agent's tool-use result, eliminating separate discovery dispatch steps.
  • Operating mode is now a mutable, schema-enforced field on the intent, allowing mid-workflow adjustments without recreating the intent.
Fixed
  • Workflow no longer gets stuck when local branches conflict with remote state during multi-stage execution.
v3.13.1
Changed
  • You can now create multiple intents in the same session without being blocked by the one-intent-per-session guard.
v3.13.0
Added
  • Run hat iterations faster with /haiku:zap, which executes a streamlined hat-loop without the usual ceremony.
v3.12.1
Fixed
  • File paths now resolve correctly when the plugin runs in worktree contexts.
  • Plugin initialization now reliably uses the bundled version when available.
v3.12.0
Added
  • Spec-conformance validation now runs automatically between execution and quality review, ensuring all stage outputs match the intent specification before quality gates assess other criteria.
v3.11.1
Fixed
  • Discovery-stage artifact merges no longer block workflow progression.
  • The Stop workflow hook now properly halts execution instead of being ignored.
v3.11.0
Added
  • Discovery templates now support model configuration, allowing you to specify which Claude version generates research artifacts and analysis for each stage.
  • Studio integrators and cross-studio decisions now support model configuration for unified governance across multiple studios.
Changed
  • Model selection now applies consistently across research, review, and decision phases throughout the workflow.
v3.10.0
Added
  • Authoring discipline instructions during the elaborate phase that clarify unit composition and feedback quality standards.
  • Fix-loop red flag guidance that identifies common quality issues in the feedback triage workflow.
v3.9.0
Added
  • Gate reviews now work in headless and remote environments with prepare/await phases, allowing review completion without a local browser UI.
  • Real-time feedback streaming via WebSocket during gate reviews for visibility in remote and CLI-only setups.
v3.8.0
Added
  • Walkthrough-oriented review with step-by-step guidance for reviewing stages and intents.
  • Per-unit output artifacts tab displays which units produced each output.
  • Declaring units banner shows output authorship at a glance.
Changed
  • Output artifacts now tracked to their declaring units for complete traceability across the workflow.
  • Git worktree handling improved for robust concurrent branch operations.
  • Tool input validation uses strict schemas to detect configuration errors earlier.
v3.7.0
Added
  • Elaborator now surfaces available skills during elaboration, and units inherit skill availability on a per-hat basis for role-specific tool access.
v3.6.1
Changed
  • Plugin binary is now bundled with the distribution, reducing installation time and improving reliability across network conditions.
  • Marketplace references now use relative paths for better offline support.
v3.6.0
Added
  • Drift detection now shows image diffs side-by-side with text changes, letting you review visual artifact modifications during code review.
  • Agents can record file modifications with attribution using haiku_record_agent_write, creating audit trails for generated content.
v3.5.0
Added
  • Drift detection alerts you when outputs change unexpectedly, with assessment interface to classify and approve changes
  • Knowledge upload during review lets you inject new files or context without restarting the intent
  • One-click output replacement for invalidated results in the review view
  • Coverage gap acknowledgment to mark sections intentionally out-of-scope
v3.4.1

No notable changes.

v3.4.0
Changed
  • The Approve button in code review now displays the consequences of your decision before you confirm
v3.3.1
Added
  • Plugin is now installable from npm
v3.3.0
Added
  • Article landmarks improve navigation and accessibility of intent, stage, and unit review content for screen readers and assistive technology.
v3.2.1
Fixed
  • Elaborate fan-out now correctly transmits parent context to discovery subagents via the prompt_file mechanism.
v3.2.0
Added
  • Plugin now displays its version in the footer and exposes version information via /api/version endpoint for programmatic access.
v3.1.0
Fixed
  • Agent now receives prior-rejection context when retrying failed units, enabling it to learn from previous attempts and avoid repeating mistakes in subsequent bolts.
v3.0.0
Changed
  • migrate now requires you to explicitly specify an intent slug or use --all instead of inferring the target.
  • migrate runs in dry-run mode by default so you can preview changes before applying them with --execute.
  • migrate will not run on a dirty working tree, preventing accidental data loss.
v2.3.0
Added
  • The plugin is now available on npm as haiku-method, simplifying installation through standard package managers.
v2.1.0
Added
  • Elaboration now includes review criteria directly in the prompt file, giving agents full visibility into review expectations before they begin working.
v2.0.0
Changed
  • Revisit workflow now routes through the feedback system, creating a unified path for addressing issues across stages.
v1.106.2
Fixed
  • The review gate no longer advances while feedback items remain open, preventing the workflow from entering conflicting states during revision cycles.
v1.106.1
Fixed
  • Fix-loop feedback resolution now properly chains when addressing one feedback item reveals or relates to additional findings.
v1.106.0
Added
  • Design-system-anchor discovery for anchoring design decisions during inception planning
Changed
  • Design stage integrated into inception phase scope
  • Design artifacts now reviewed to ensure they are properly grounded in inception planning decisions
v1.105.25
Fixed
  • Review displays all stage artifacts, including nested files and unrecognized file types.
v1.105.23
Fixed
  • Worktrees are now properly isolated from version control and won't be accidentally included in commits.
v1.105.22
Fixed
  • Unit field updates now include type validation and schema hints, giving agents clearer error messages when setting invalid field values.
v1.105.21
Fixed
  • Plugin now reliably launches the browser on Windows and correctly detects its own installation directory across platforms.
v1.105.20
Fixed
  • Design direction selections now persist if the design picker tool is cancelled, preventing loss of your screenshot choices.
v1.105.19
Fixed
  • Feedback dispatch now reliably flows through a centralized pre-tick triage gate, ensuring consistent routing of human-authored feedback items.
v1.105.18
Fixed
  • The review UI no longer reappears for feedback left unresolved; unaddressed findings are now routed back to the agent for inline triage.
Changed
  • Studio architecture diagrams are now native Next.js components, improving website performance and maintainability.
v1.105.17
Fixed
  • State tracking is now unified across the orchestrator, API, and browser interface, ensuring consistent intent status across all entry points.
v1.105.14
Changed
  • Aligned session UIs (direction and question pages) to the review interface's annotation pattern for consistent feedback display and decision tracking across all workflow stages.
v1.105.13
Fixed
  • Intent creation in git repositories now correctly branches from the mainline reference without modifying the active checkout state.
v1.105.12
Added
  • New MCP tools enabling programmatic management of units and feedback items throughout their lifecycle.
  • Path-boundary protection ensuring that FSM-managed files cannot be accidentally modified by agents or scripts.
  • Canonical ARCHITECTURE.md reference documenting studio, stage, hat, unit, and feedback structures.
Changed
  • Fix loops now dispatch review feedback through stage-level hat sequences with FSM-enforced lifecycle progression.
  • Elaboration-phase criteria explicitly defined per stage in phases/ELABORATION.md files.
Removed
  • Studio-level reconciler and validator fix hats, replaced by stage-level feedback-driven fix dispatch.
v1.105.10
Added
  • Quality gate enforcement: stages block advancement until quality criteria are verified.
  • Elaboration phase now captures structured decision reasoning across all studio types.
Changed
  • Review agents now enforce domain-specific verification requirements at stage gates.
v1.105.9
Added
  • Hardware design workflows now support PCB layout and schematic capture through tscircuit EDA platform integration.
v1.105.8
Added
  • Conversation examples for each H·AI·K·U studio demonstrating how to apply the methodology across different domains and workflows.
v1.105.7
Added
  • Programmatic feedback API with OpenAPI specification for integrating H·AI·K·U feedback with external systems and automating recovery workflows
  • Studio-level review agents that automatically enforce consistency across stages without manual intervention
  • Autonomous feedback assessor in every stage for independent closure decisions instead of manual sign-off
  • Standalone review UI package for independent deployment and flexible tool integration
Fixed
  • Tailwind v4 build compatibility and TypeScript compilation errors in CI pipeline
v1.105.6
Fixed
  • Enhanced error observability throughout the orchestrator for improved debugging and diagnostics.
  • Hardened intent and unit metadata parsing to prevent edge-case processing failures.
v1.105.5
Added
  • OpenTelemetry integration for comprehensive workflow observability
  • Authentication helpers for credential management
Fixed
  • Browse operations now recover gracefully from transient network failures
v1.105.4
Fixed
  • Discrete mode no longer creates a merge request for each unit.
v1.105.3
Added
  • Support for running H·AI·K·U in Cursor, Gemini CLI, Kiro, OpenCode, and Windsurf.
v1.105.2
Added
  • Review sessions preserve draft responses so you can continue editing incomplete reviews later.
  • Review re-entry shows what changed since your last review to accelerate feedback iteration.
Fixed
  • Eliminated duplicate confirmation popups during review updates.
v1.105.1
Fixed
  • Execution phase no longer over-prompts during unit iteration, letting you progress through work without unnecessary interruptions.
v1.103.2
Fixed
  • Repair tool no longer gets stuck in stale-reference loops and now respects your repository's actual mainline branch instead of hardcoding assumptions.
v1.103.0
Added
  • Comprehensive artifact injection across all FSM phases, allowing agents to reference prior discoveries, operations, reflections, and outputs throughout the entire workflow.
v1.102.2
Changed
  • Agents now receive improved context when formulating questions during orchestration, resulting in clearer agent-to-agent interactions.
v1.102.1
Fixed
  • Stages with review: auto gates now automatically advance without waiting for user input.
v1.102.0
Added
  • Automatic repair of migrated intents with incomplete prior stages, allowing safe resumption of work without data loss
v1.101.7
Fixed
  • External review gates now reliably coordinate with external review systems for seamless stage advancement.
v1.101.6
Fixed
  • Remote code review now works end-to-end with stable HTTP connections using heartbeat mechanism to keep sessions alive
v1.101.4
Added
  • Example intents for hardware development, content creation, incident response, legal review, and OAuth implementation workflows
Changed
  • Improved Mermaid diagram rendering for scavenger hunt feature visualization
v1.101.3
Fixed
  • Sentry error monitoring is now properly embedded in the plugin binary and reliably captures errors in production.
v1.101.1
Fixed
  • Directory-type artifact locations are now properly resolved in stage inputs.
v1.101.0
Added
  • MCP binary now automatically updates from GitHub releases without requiring manual intervention.
v1.100.1
Added
  • Mermaid flowcharts now render as interactive React Flow diagrams you can explore, pan, and zoom.
  • Automatic detection and handling of discrete commits and squash merges on hub branches for improved state tracking.
Fixed
  • Removed references to deleted unit.type frontmatter field from review application.
v1.100.0
Added
  • Revisit command to pick up where you left off on previous work and intents
Changed
  • Repair command now handles migrated intents and edge cases more reliably
v1.99.0
Added
  • Discrete branching with three-level trust merge in browse for safer, more granular code exploration and integration.
v1.98.1
Fixed
  • Archived intents stored only on the mainline branch now properly load and are available for work.
v1.98.0
Added
  • Game Development studio for interactive products with concept, prototype, polish, production, and release stages — includes playtesting validation and scope discipline gates
  • Hardware Development studio for physical products with design, firmware, requirements, manufacturing, and validation stages — includes compliance mapping and manufacturability review
  • Library Development studio for APIs and packages with inception, development, security, and release stages — includes semantic versioning enforcement and API stability gates
v1.97.3
Fixed
  • Output validation now properly enforces that stage outputs cannot be empty, preventing invalid state progression.
v1.97.2
Added
  • Cascading model selection: specify which Claude model to use at different workflow levels (studio, stage, or unit) with automatic inheritance from parent scopes
v1.97.1
Fixed
  • Stage inputs now correctly resolve from previous stage outputs, enabling reliable data flow between stages
  • Unit inputs are tracked and enforced throughout execution, preventing silent loss of input context
v1.97.0
Added
  • Run-next responses now preview the next step and upcoming actions before execution.
v1.96.0
Added
  • Per-stage git branching with discrete mode (isolated branch per stage) or hybrid mode (customizable mix of shared and isolated stages) for flexible workflow organization.
v1.95.4
Changed
  • Product stage now separates product ownership, specification writing, and validation into three independent roles for more granular control of product definition workflows.
v1.95.3
Added
  • New /haiku:report skill for generating structured project reports
Changed
  • Renamed skills for clarity: /haiku:release-notes → /haiku:changelog, /haiku:review → /haiku:gate-review, /haiku:resume → /haiku:pickup
Fixed
  • Error telemetry now correctly routes to self-hosted Sentry instance
v1.95.2
Fixed
  • Removed outdated references to get_review_status tool from validation documentation.
v1.95.1
Added
  • Share remote review sessions with collaborators using localtunnel.
Fixed
  • Intent review gate now correctly enforces review stage gates.
  • Intent review workflow handles tunnel reconnections during code change review.
v1.95.0
Added
  • H·AI·K·U prompts now available as Claude Code skills, enabling use in cowork collaborative sessions
v1.94.0
Added
  • Quality gates are now automatically enforced during orchestration, validating completion criteria before allowing progress.
Fixed
  • Intent and unit review interfaces now have improved layout and usability.
v1.93.1
Changed
  • Added comprehensive test suite for MCP handlers, orchestration logic, and core utilities to ensure plugin reliability and maintainability.
  • ---
  • Note: This patch version contains only internal test infrastructure improvements with no user-facing changes.
v1.93.0
Added
  • Researcher hat for inception stage that aligns discovery and problem-solving with product expectations.
Removed
  • Architect and decomposer hats from inception stage.
Changed
  • Discovery phase in inception stage updated to incorporate researcher workflows.
v1.92.0
Added
  • Trace artifact outputs back to their source units with automatic FSM validation, ensuring consistency across your project's generated work.
v1.91.0
Added
  • Intent details now display full content including knowledge, artifacts, and designs.
v1.89.1
Fixed
  • Parallel execution prompts now provide more forceful guidance, and inception stage maintains business-level focus throughout planning.
v1.89.0
Added
  • Plugin now automatically detects your project type (git repository vs filesystem) and configures persistence accordingly, eliminating manual setup
Changed
  • Tool orchestration is now driven by a finite state machine, providing more reliable and predictable execution
  • Studio configurations are simplified with automatic persistence and state management setup
v1.88.0
Added
  • Git provider integration enables remote repository browsing and file operations from the plugin.
Changed
  • Orchestrator execution engine refactored for more reliable stage transitions and state management.
  • Review sidebar redesigned to improve navigation and context visibility during unit and gate reviews.
Fixed
  • Validation catches configuration errors more reliably, preventing invalid states from progressing through stages.
v1.87.6
Fixed
  • Unit naming validation now runs before review across all stages, preventing naming inconsistencies from reaching the review phase.
v1.87.5
Fixed
  • Studio execution now correctly detects unresolved depends_on references instead of falsely reporting circular dependency errors.
v1.87.4
Fixed
  • Resolved circular dependency issues in core plugin initialization
v1.87.3
Fixed
  • Review workflow is now consistently enforced through the state machine, ensuring review gates cannot be bypassed by direct initiation.
v1.87.2
Added
  • Design annotations on unit review pages to mark component boundaries and interactions visually.
Changed
  • Renamed /haiku:new command to /haiku:start for creating new intents.
Fixed
  • Review sidebar now stays visible when scrolling through unit and gate details.
  • Unit naming now validates against H·AI·K·U conventions to prevent invalid slugs.
Removed
  • Plugin skills system (replaced by MCP prompts for more reliable orchestration).
v1.87.1
Fixed
  • Auth proxy load balancer now uses Google-managed certificates for improved security and reliability.
v1.87.0
Changed
  • Resume work on existing intents with haiku:resume (formerly haiku:run), and start new work with a simplified haiku:start command.
v1.86.11
Fixed
  • Unit parsing now strictly enforces proper structure, removing incorrect fallback behavior that could mask malformed units.
v1.86.10
Fixed
  • Fixed proxy-only subnet CIDR conflicts and added proper infrastructure deployment dependencies.
v1.86.7
Fixed
  • Fixed infrastructure configuration to enable proper plugin deployment
v1.86.5
Fixed
  • Terraform deployments can now recover from partial applies by importing orphaned GCP resources instead of failing.
v1.86.4
Fixed
  • Improved authentication proxy load balancing for better reliability under load.
v1.86.3
Fixed
  • Authentication proxy deployments now correctly reference serverless Network Endpoint Groups by service name instead of full path, preventing configuration errors in serverless environments
v1.86.2
Fixed
  • Backend service now properly scales capacity based on demand, improving reliability under load.
v1.86.0
Added
  • Regional HTTPS load balancer support for auth proxy public access, improving global availability and performance.
v1.85.5
Fixed
  • Auth proxy URL now updates automatically during deployments.
v1.85.4
Fixed
  • Deployments no longer fail when stale Terraform locks block the apply step.
v1.85.3
Fixed
  • Fixed authentication proxy to use ingress settings for public access instead of IAM roles.
v1.85.2

No notable changes.

v1.85.1
Fixed
  • Tailwindcss binary is now resolved via module resolution instead of hardcoded paths, improving compatibility across different environments.
  • Workspace configuration updated to properly include packages/shared in monorepo builds.
  • Based on your commits, this could be expanded to mention the major feature (FSM orchestrator + review UI from PR #144) and infrastructure improvements. Let me know which version you'd like me to focus on, and I'll write a more comprehensive changelog entry.
v1.85.5
Fixed
  • Auth proxy URL repository variable is now automatically set from Terraform output during deployment.
v1.85.3
Fixed
  • Plugin now automatically detects your GCP project from service account keys, eliminating manual configuration of the project ID and state bucket.
v1.85.2
Fixed
  • Plugin deployment now works correctly when the repository is forked.
v1.85.1
Fixed
  • Tailwindcss binary is now resolved via module resolution instead of hardcoded paths, improving compatibility across different environments.
  • Workspace configuration updated to properly include packages/shared in monorepo builds.
v1.85.0

No notable changes.

v1.84.1
Fixed
  • Reverted haiku rebrand to previous state
  • If you'd like me to check the actual diff to provide more specific detail about what was reverted, I can read the current state compared to the previous release. Would you like me to do that?
v1.84.0
Added
  • Design provider support — reference design artifacts from Figma, Canva, Excalidraw, OpenPencil, Pencil, and Penpot; the plugin auto-detects and integrates them into your visual review workflows
  • Post-integrate knowledge refresh — knowledge context automatically updates after integration runs
  • Native design provider link resolution — handles both web URLs and native provider links for seamless artifact reference
Changed
  • Designer hat now leverages design provider artifacts in visual reviews and wireframe elaboration for context-aware feedback
v1.83.0
Added
  • Subscribe to changelog and blog updates via RSS, Atom, and JSON feeds
  • Release-notes skill for accessing release information
Fixed
  • Feed generation now properly escapes URLs in XML and formats Atom dates correctly
v1.82.13
Changed
  • Reorganized internal subskills to nest under their parent skill directories for clearer plugin structure.
Fixed
  • Corrected stale command references and paths throughout skill definitions and artifacts to ensure proper subskill invocation.
v1.82.12
Fixed
  • Agent teams created during a unit are now properly cleaned up when the unit completes.
v1.82.11
Changed
  • Changelog generation is now AI-synthesized for consistency across release notes.
  • But strictly speaking, this is a tooling-only patch with no user-facing changes.
v1.83.1
Fixed
  • Reverted Visual Review full integration pending further development.
v1.82.8
Added
  • /ai-dlc:review skill for pre-delivery code review with automated fix loop, integrated into quick mode
Changed
  • Automatic setup now runs if settings.yml is missing
  • Quality gates handling improved with clearer formatting and support for 5+ gates
  • Elaboration now less verbose with simplified setup concerns
  • PR terminology standardized across skills and documentation
v1.82.7
Fixed
  • Plugin installation is now more reliable and no longer requires npm
  • CPU architecture is automatically detected for cross-platform binary dependency support
v1.82.6
Added
  • Knowledge synthesis and design direction system with archetype library and visual picker for elaboration bootstrapping
  • Three new elaboration phases: 2.3 (Knowledge Bootstrap), 2.75 (Design Direction), 7.5 (Adversarial Review), plus autonomous elaboration mode for /autopilot
  • /adopt skill to reverse-engineer existing code into AI-DLC intent artifacts
  • Dynamic pass definitions allowing custom passes in construction pipeline
  • Quality gates with visual review interface, intent dashboard, and mermaid-enabled MCP server
  • BDD workflow with acceptance-test-writer hat for behavior-driven elaboration
Removed
  • workflow_mode setting — /autopilot is now autonomous by definition
v1.82.5

No notable changes.

v1.82.4
Added
  • Visual-review intent to structure collaborative design and UX review workflows
  • Sync rules documenting consistency requirements between paper, plugin, and website components
v1.82.3
Fixed
  • Intent status is now validated before delivery to catch configuration issues early.
v1.82.2
Fixed
  • Intent status now correctly updates when Teams path completes
Removed
  • Standalone HTML guides (content now available on website)
v1.82.1
Fixed
  • Planner hat no longer incorrectly enters plan mode, strategy options are properly ordered, and intent strategy no longer prompts for auto-merge confirmation.
v1.82.0
Added
  • Story-driven home page and plugin lifecycle guides provide intuitive introduction to the AI-DLC methodology and workflow.
v1.81.1
Fixed
  • Intent strategy is now the default workflow, eliminating the auto-merge confirmation prompt.
v1.81.0
Added
  • Passes are now configurable, letting you customize which disciplinary lenses (design, product, dev) iterate in your workflows.
Changed
  • Construction Phase is now called Execution Phase.
v1.80.1
Fixed
  • Plugin hooks no longer depend on external han library, reducing potential failure points and improving reliability.
v1.80.0
Added
  • Harness framework documentation in the methodology paper and website, explaining how the plugin execution model works.
v1.79.2
Fixed
  • Removed outdated han-plugin references from skill documentation.
v1.79.1
Fixed
  • Advance skill now properly evaluates completion gates against the actual plan file
  • Resume skill correctly parses YAML frontmatter in unit specifications
  • Haiku utility can now be called standalone with automatic state dependency resolution
v1.79.0
Added
  • Infrastructure and operations as a workflow discipline: Elaborate now guides you through infrastructure questions and auto-creates operations units alongside feature work
  • Operations construction phase that builds and validates deployment and infrastructure configurations during the execution flow
  • Operational readiness reviews that validate infrastructure and deployment completeness before shipping
  • Stack configuration schema with quality gates for defining infrastructure patterns and architectural standards
v1.78.1
Fixed
  • Fixed visual-fidelity issues identified in code review
v1.78.0
Added
  • Completion criteria checkboxes automatically check off when a unit or intent is completed, keeping your workflow state synchronized with task progress.
v1.77.2
Fixed
  • Interactive workflow mode now has write permissions enabled.
v1.77.1
Fixed
  • Code review workflow now has the necessary permissions to write and apply changes when reviewing code.
v1.77.0
Added
  • Visual fidelity review agent with gate detection and comparison orchestration for automated design quality assessment, including backpressure documentation for integration workflows
  • Design reference resolver that intelligently locates design system components using a 3-level priority hierarchy
  • Pluggable screenshot capture infrastructure enabling flexible image acquisition across different design tools
Fixed
  • Prevent stderr/JSON mixing in design reference resolution calls
v1.76.2
Fixed
  • Intent and unit completion statuses are now reliably set when work finishes.
v1.76.1
Changed
  • Plugin no longer depends on han CLI; core state management and hook utilities are now built into the plugin via foundation libraries
  • Simplified iteration.json state schema with improved phase validation
v1.76.0
Added
  • Enforce quality checks with hard gates that must pass before unit completion.
  • Automatically delegate work to specialized subagents using configurable hooks.
  • Generate DOT flowcharts to visualize intent and unit workflows.
  • Validate execution context with preflight checks before running workflows.
v1.75.1

No notable changes.

v1.75.0
Added
  • Code reviews now support multi-judge critique debate for high-stakes decisions.
v1.74.2
Fixed
  • Fixed CI pipeline conflict resolution to properly handle push events.
v1.74.1
Fixed
  • Context:fork skill invocations during elaboration now use the Agent tool for improved execution reliability
v1.74.0
Added
  • Schema drift detection for database changes during code review
v1.73.0
Added
  • Use /autopilot skill to run complete autonomous feature lifecycles from elaboration through delivery.
Changed
  • /execute is now the primary execution skill; /construct is deprecated.
v1.72.0
Added
  • Reviewers can now use structured completion markers to mark deterministic review outcomes with confidence-based feedback.
v1.71.0
Added
  • Builder includes a node repair operator and structured completion markers for improved construction reliability and state tracking.
v1.70.1
Fixed
  • Improved conflict resolution in continuous integration builds using matrix strategy.
v1.70.0
Added
  • The planner now analyzes git history to provide more informed planning recommendations.
v1.68.1
Fixed
  • Shell scripts are now executable after installation, eliminating permission errors.
Added
  • Workflow mode selection and granularity tuning options to customize AI-DLC iteration behavior.
  • Documentation for integrating Reflect into the Compound skill workflow.
v1.68.0
Added
  • DOT flowchart guidelines for visualizing process authority during the construction phase.
v1.67.0
Added
  • Model profiles for cost-optimized hat routing with improved configuration resolution
v1.66.0
Added
  • Planning process now retrieves relevant learnings before generating plans, with improved search capabilities
Fixed
  • Restored Structured Completion Marker section that was accidentally removed
v1.64.0
Fixed
  • Restored missing Two-Stage Review, conflict resolution, CoVe, and Specialized Pre-Delivery Reviews sections in reviewer hat
Added
  • Parallel review perspectives for multi-file units enable concurrent reviews from multiple disciplinary angles
v1.63.0
Added
  • Three-level artifact checking in reviews validates content, reasoning, and structural integrity
  • Goal-backward verification ensures work aligns with original intent throughout development
  • Anti-rationalization tables and red flags across all hats identify biased reasoning patterns
  • Confidence scoring system enables granular quality assessment during review cycles
  • Enhanced anti-pattern detection and specialized review modes for deeper code analysis
v1.62.3
Fixed
  • Conflict resolution in automated workflows now requests manual review instead of taking automatic action, improving reliability when handling merge conflicts.
v1.62.2
Changed
  • Documentation terminology updated throughout to reflect current plugin naming, removing references to legacy terms.
v1.62.1
Fixed
  • Conflict resolver polling now operates correctly during workflow execution.
v1.62.0
Added
  • Customize elaboration phases for your project using .ai-dlc/ELABORATION.md configuration files
  • Configure external rule files using a data-driven pattern
v1.61.1

No notable changes.

v1.61.0
Changed
  • Reviewer hat now validates spec compliance first (including test coverage requirements), then assesses code quality in a separate stage.
v1.60.0
Added
  • Visual brainstorming companion guidance during the elaboration phase
Fixed
  • Brainstorm storage path to prevent collisions with mockups directory
v1.59.1

No notable changes.

v1.59.0
Added
  • Context budget monitoring warns when context usage approaches limits at 35% and 25% remaining, helping you manage token usage during long sessions.
v1.58.2
Changed
  • Builder now reduces token consumption by approximately 35% by extracting reference materials to a companion file.
v1.58.1
Changed
  • Reduced token consumption during code reviews by extracting reference material to a companion file.
v1.58.0
Added
  • Structured completion markers explicitly control when the workflow transitions between hats, replacing implicit transition logic with deterministic gating.
v1.57.0
Added
  • Code reviews now include confidence-scored findings and automated anti-pattern detection.
v1.56.1
Changed
  • Session token consumption is reduced through optimized hook lazy loading
v1.56.0
Added
  • Capture learnings from the current session as structured solution files with the new /compound skill.
v1.55.0
Added
  • Use anti-patterns guidance when writing completion criteria to avoid common mistakes that over-constrain or inappropriately restrict agent output.
v1.54.0
Added
  • Added /pressure-testing skill for test-driven development with quality gate validation.
v1.53.1
Added
  • You can now disable model invocation in the reflect, resume, and operate skills to prevent unintended model calls during execution.
v1.53.0
Added
  • Execute multiple units within an intent in parallel waves, automatically coordinating dependencies and sequencing units for optimal throughput.
v1.52.0
Added
  • Sessions now resume with structured handoff to maintain context and progress across bolts.
v1.51.2

No notable changes.

v1.51.1
Changed
  • Review operations now consume fewer tokens due to optimized subagent context scoping
v1.51.0
Added
  • Discovery phase now supports brownfield codebase mapping to help you understand existing code architecture when elaborating new intents.
v1.50.0
Added
  • Aggregate learnings from compound work into structured reflection outputs for project documentation and review.
v1.49.0
Added
  • Workflows now enforce quality gates at hard-gate synchronization points during critical phase transitions.
v1.48.0
Added
  • Elaboration now includes automated specification review before advancing to construction.
v1.47.0
Added
  • Builder now requires verification of completion criteria before marking units complete.
v1.46.0
Added
  • Use the /seed skill to capture forward-looking ideas and define conditions for when they should be revisited.
v1.45.0
Changed
  • Reviewer now uses specialized review agents to provide comprehensive pre-delivery checks instead of a static checklist
v1.44.3
Added
  • Git strategy questions now show your actual default branch name, and you can configure which remote to use for detection.
v1.44.2
Added
  • Claude Code Review workflow for conducting code reviews directly within Claude Code.
Changed
  • Improved Claude PR Assistant workflow with enhanced pull request management capabilities.
v1.44.1

No notable changes.

v1.44.0
Added
  • Plugin state now persists to disk with lockfile protection, enabling safe multi-session workflows.
v1.43.0
Added
  • Safely execute untrusted or generated prompts with built-in prompt injection guards
  • Configure workflow enforcement hooks to control tool execution and maintain process discipline
v1.42.0
Added
  • Use /quick mode to skip full elaboration for trivial tasks.
v1.41.0
Added
  • Agent teams can now use a parallel commit strategy with no-verify enabled to speed up concurrent builds.
v1.40.0
Added
  • Specialized review agents organized by domain, enabling more targeted feedback based on codebase expertise.
v1.39.0
Added
  • You can now use /backlog to capture and organize parking lot ideas during development.
v1.38.0
Added
  • Generate improvement suggestions with adversarial review using the new /ideate skill.
v1.37.0
Added
  • Run parallel research agents to deepen plans and explore multiple angles simultaneously
v1.36.0
Added
  • Spec flow analysis during elaboration helps identify specification gaps before implementation.
v1.35.0
Added
  • Customize review agent configuration on a per-project basis to align code review behavior with your team's standards.
v1.34.0
Added
  • Reviewer now uses chain-of-verification (CoVe) methodology to provide evidence-based code reviews grounded in systematic analysis.
v1.33.0
Added
  • Search for learning materials in the planner and get results ranked by relevance using multiple signals to quickly find applicable guidance.
v1.32.0
Added
  • Builder automatically detects version incompatibilities and provides rollback guidance during builds.
v1.31.0
Added
  • The planner now uses rule-based filtering to automatically select the best approach for your task.
v1.30.0
Added
  • Configuration documentation now explains the precedence pattern for master and override settings, allowing you to correctly configure layered settings.
v1.29.0
Added
  • Units now include a last_updated timestamp in their frontmatter, allowing you to track when each unit was last modified.
v1.28.2
Fixed
  • Unit and intent status changes are now committed to git automatically, ensuring all workflow state transitions are tracked in version control.
v1.28.1
Changed
  • Strategy options documentation now explains each choice from the user's perspective, making it clearer when to use each strategy.
v1.28.0
Added
  • Automatic cleanup of temporary git worktrees at completion milestones, eliminating manual cleanup steps after intent completion.
v1.27.0
Added
  • Design-specific unit template sections to guide structured documentation of design requirements during intent elaboration.
v1.26.0
Changed
  • /followup now creates linked iteration intents to maintain continuity across intent iterations.
v1.25.0
Added
  • Wireframes now include design units during the elaboration phase for more complete design specifications.
v1.24.0
Added
  • Receive design-focused guidance when defining success criteria during the elaborate phase.
v1.23.0
Added
  • Design discipline now automatically routes to a design-focused workflow.
v1.22.0
Added
  • Plugin now automatically detects and removes worktrees once their changes are merged.
v1.21.0
Added
  • Workflow execution events are now reported to OpenTelemetry, enabling integration with observability platforms and monitoring tools.
v1.20.16
Changed
  • Documentation rebrand with no user-facing changes
v1.20.15
Added
  • Test-driven development workflow for building features from tests first
  • Learnings from completed intents are automatically loaded into new workflows
Changed
  • Integration workflows now run autonomously without manual intervention
v1.20.14
Removed
  • Operator and Reflector hats and their associated workflows have been removed from the available workflow options.
v1.20.13
Fixed
  • Plugin can now be properly discovered and installed via the Claude Code marketplace
v1.20.12
Fixed
  • Unit strategy automations no longer ask for auto-merge confirmation during setup.
Changed
  • Simplified plugin source configuration.
v1.20.11
Changed
  • Restructured plugin directory and marketplace configuration
v1.20.10
Added
  • /reflect skill for reflection phase—review and learn from completed work
  • /operate skill for operation phase—maintain and iterate on deployed systems
  • AI-DLC integrates with HAIKU software development methodology
Changed
  • Construction phase renamed to Execution throughout plugin and documentation
v1.20.9
Added
  • Plugin can be downloaded as a zip archive
Changed
  • Marketplace now uses path-based source configuration
v1.20.8
Changed
  • H·AI·K·U is now discoverable in the plugin marketplace with full description
v1.20.7

No notable changes.

v1.20.6
Fixed
  • Plugin now correctly resolves worktree paths relative to the main repository root instead of the worktree directory, preventing path resolution errors when using git worktrees.
v1.20.5
Fixed
  • Review pull requests created during elaboration no longer close linked issues when merged.
v1.20.4
Fixed
  • Unit-based change strategy no longer prompts for delivery confirmation, streamlining the workflow.
v1.20.3
Fixed
  • Elaboration review now displays complete unit content without truncation.
v1.20.1
Fixed
  • Discovery skill now enforces strict alignment rules for ASCII wireframes, improving consistency and accuracy during domain exploration.
v1.20.0
Added
  • Elaborate phases can now be run independently as fork subagent skills, giving you granular control over which phases to execute.
v1.19.2
Fixed
  • Worktree directories are now automatically added to gitignore, preventing temporary work from being accidentally committed.
v1.19.1
Fixed
  • Elaboration workflow is now unified, removing the complexity of managing separate elaborator agent and elaboration-start skill components.
v1.19.0
Added
  • Changelog page on the website for viewing release history
Changed
  • Elaborate workflow now orchestrates multiple agents for better intent specification
v1.18.0
Added
  • Detect greenfield projects and automatically generate UI mockups for rapid design exploration.
v1.17.2
Fixed
  • Changelog generation now includes only commits since the previous version for more accurate release notes
v1.17.1
Fixed
  • Intent discovery now creates a worktree before running, preventing temporary artifacts from appearing on the main branch.
v1.17.0
Added
  • Agents can now invoke elaborate, resume, and refine skills, enabling automated AI-DLC workflows.
v1.16.0
Added
  • Discovery scratchpad automatically captures elaboration findings to discovery.md during domain discovery, reducing context window pressure while keeping detailed findings accessible to subagents and the intent worktree
  • Parallel design analysis via dedicated subagents during elaboration for faster, context-efficient Figma and design file processing
  • Hybrid per-unit change strategy support, allowing mixed strategies within a single intent—foundational units can use direct-to-main while remaining units merge to the intent branch
v1.15.0
Added
  • Per-unit workflows enable different workflow configurations for each unit, with built-in support for design discipline phases.
v1.14.0
Added
  • Handle design assets directly in your workflows for easier design integration
  • Automatically match and apply colors from design specifications
  • Leverage design annotations to guide code generation and specifications
v1.13.0
Added
  • Handoff now supports cowork-aware modes with options to share work as a local folder or zip file for easier collaboration.
v1.12.0
Changed
  • Slash commands /reset, /refine, /setup, and /resume are now blocked in cowork mode to prevent conflicting simultaneous operations.
v1.11.0
Changed
  • The /construct command is now unavailable in cowork mode to prevent mode conflicts during collaborative work.
v1.10.0
Added
  • Cowork mode now automatically detects multi-agent collaboration environments, improving coordination when multiple Claude Code instances work together.
  • Exploration tasks now delegate to specialized Explore subagents for faster codebase discovery and analysis.
v1.9.0
Added
  • Wireframe generation phase for visual design exploration and iteration in development workflows
Changed
  • Worktrees are now automatically created within project directories for better isolation and organization
v1.8.3
Fixed
  • Ticket descriptions now display with improved formatting and structure for better readability.
v1.8.2
Changed
  • Reference skills are now internal-only and cannot be invoked directly
v1.8.1

No notable changes.

v1.8.0
Added
  • Execute specific units without running the entire intent
Changed
  • Change strategies now provide more options for managing workflow iterations
Removed
  • Bolt change strategy has been removed; use the intent strategy instead
v1.7.0
Added
  • Completion announcements signal when tasks meet their completion criteria.
  • Risk descriptions let you document potential issues identified during elaboration.
  • Iteration caps set a maximum limit on how many times a bolt can cycle.
  • Explicit bolt terminology clarifies the iteration cycle concept throughout the plugin.
v1.6.3
Changed
  • Simplified skill definitions and hook configurations by removing mode references
v1.6.2
Removed
  • Mode selection from elaboration and construction phases.
v1.6.1
Fixed
  • Iteration state is now correctly initialized during construction instead of elaboration, ensuring proper workflow sequencing from project inception through delivery.
v1.6.0
Added
  • Integrator hat: Validate cross-cutting concerns after merging all units within an intent.
  • /refine skill: Refine intent or unit specifications mid-execution without losing progress.
  • Non-functional requirement guidance: Enhanced prompts to help you specify and validate non-functional requirements and cross-cutting concerns.
  • Delivery phase guidance: Improved prompts for planning and executing the delivery phase.
v1.5.0
Added
  • New /setup skill automates initial project configuration.
  • Elaboration now enforces ticket creation for all work before execution can begin.
v1.4.5
Fixed
  • Testing requirements are now mandatory for all intents and cannot be configured per-intent.
v1.4.4
Fixed
  • Subagents now load context state from the correct branch when executing work.
v1.4.3
Fixed
  • Intent branches are now prefixed with ai-dlc/ to prevent naming conflicts and improve branch organization.
v1.4.2
Fixed
  • Construction workflows now skip the elaborator phase and intent discovery has been improved.
v1.4.1
Fixed
  • Plugin installation now uses Claude Code's native /plugin CLI for streamlined setup.
v1.4.0
Added
  • Integrate external providers to customize how agents access memory, data, and external services
  • Enable coworking mode for team-based and collaborative AI-DLC workflows
  • Instructions now merge hierarchically across global, project, and local levels so local customizations take precedence
v1.3.0
Added
  • Configure memory providers to customize where and how AI-DLC state is stored.
  • Collaborate with team members through cowork support for multi-user development workflows.
Changed
  • Plugin structure reorganized for improved modularity and maintainability.
v1.2.2
Changed
  • Intent configuration is now consolidated into intent.md frontmatter, eliminating the need for a separate intent.yaml file.
v1.2.1
Fixed
  • Session retrospectives now correctly handle branch ordering and workflow transitions
  • Team mode hats now function properly in multi-agent workflows
  • Merge strategy now executes as configured
v1.2.0
Added
  • Domain discovery during elaboration helps you systematically explore your problem space before constructing solutions
  • Specification validation during elaboration catches design issues early in the inception phase
  • Deep research capabilities help you gather comprehensive context about relevant technologies and approaches
v1.1.2

No notable changes.

v1.1.1
Added
  • Agent Teams with intent-level modes and dynamic hat discovery
  • Interactive workflow visualizer on documentation website
  • Website responsive layout and dark mode
Changed
  • Construction workflows now use Agent Teams by default
Fixed
  • Session startup performance optimized