GitHub Releases

Changelog that follows the repo.

Release notes are pulled from GitHub Releases, then grouped by their Markdown sections so the public changelog stays aligned with what ships.

Subscribe or inspect the source.
Subscribe via RSSBrowse GitHub releases

v0.1.16

GitHub

Patch release for Memory Center Dream scheduling. This keeps the v0.1.15 Memory Center work and fixes the accepted Dream proposal path so the Dream page actually becomes scheduled.

Highlights

  • Fix Memory Center Dream scheduling proposals so accepting a side-chat dream-dry-run proposal writes the Dream scheduler state instead of only saving an ordinary memory proposal.
  • Recover Dream scheduler state from already-applied Dream proposals, so users who accepted a Dream schedule proposal in 0.1.15 see the Dream page become scheduled after upgrading.
  • Add Dream schedule parsing for human time windows such as 6pm a 11pm en Panama, 18:00-23:00, and fixed times such as 21:00 America/Panama.
  • Add regression coverage for side-chat Dream proposal application, Dream schedule recovery, and human-readable Dream time parsing.

Added

  • Add Dream schedule parsing for human time windows such as 6pm a 11pm en Panama, 18:00-23:00, and fixed times such as 21:00 America/Panama.
  • Add regression coverage for side-chat Dream proposal application, Dream schedule recovery, and human-readable Dream time parsing.

Fixed

  • Fix Memory Center Dream scheduling proposals so accepting a side-chat dream-dry-run proposal writes the Dream scheduler state instead of only saving an ordinary memory proposal.
  • Recover Dream scheduler state from already-applied Dream proposals, so users who accepted a Dream schedule proposal in 0.1.15 see the Dream page become scheduled after upgrading.

v0.1.15

GitHub

Large Memory Center and terminal UX release. It adds the route-level memory workspace, Dream foundations, side-chat proposal flow, richer docs/screenshots, prompt/model metadata fixes, Markdown/Plan rendering upgrades, Herdr state integration, project MCP config, and release dependency hardening.

Highlights

  • Memory Center: Overview, Project memories, Global memories, Policy & categories, Dream, proposal review, and side chat.
  • Dream foundations: bounded evidence collection, logs, scheduler state, source manifests, missed-window handling, and safety reporting.
  • Terminal UX: Markdown/Plan rendering, Mermaid support, chat timeline polish, prompt chrome fixes, and Herdr agent-state reporting.
  • Workflow/runtime: project MCP config sync, package/setup coverage, worktree/mflow control-plane polish, and release dependency hardening.

Added

  • Add a route-level Memory Center in the TUI with Overview, Project memories, Global memories, Policy & categories, and Dream tabs.
  • Add saved/pending memory review surfaces with keyboard actions for applying, rejecting, editing, deleting, moving, and inspecting memories without leaving the terminal.
  • Add category-aware memory metadata, policy normalization, scope reasoning, evidence references, and reviewable proposal operations for verify, expire, recategorize, relink, demote-scope, and promote-scope flows.
  • Add a memory graph sidecar with legacy-entry bridging so memories can be grouped, indexed, and queried by category while preserving the existing flat memory files.
  • Add workspace-aware memory grouping so the Memory Center can show the current project plus other known project memory scopes.

Fixed

  • Make the prompt model, provider, and variant labels resolve from the selected prompt model metadata instead of stale local/session labels.
  • Preserve assistant provider/model/variant metadata when sync events arrive with partial message updates.
  • Keep prompt status script output tied to the current prompt-status identity so stale script output from an older model/session/root is not reused.
  • Replace the Plan Review renderer path with a stable styled Markdown component instead of relying on the experimental Markdown flag.

Changed

  • Label retrieved prompt memory by scope and category while preserving the existing <mendcode_memory> compatibility block.
  • Feed recent conversation context into post-turn memory extraction so proposals can distinguish durable facts from isolated one-line messages.
  • Track whether local model and variant state came from the user or session hydration, so explicit local changes win over older hydrated session state.
  • Persist submitted prompt agent/model selections onto the session before generation so resumed sessions and Agent View reflect the real active selection.
  • Improve Plan Review table rendering with grid mode, wrapped cells, and better handling for file/action/detail tables.

Docs

  • Add screenshot-backed public docs for the README hero, home layouts, Agent View, Plan Mode, and Memory Center.
  • Add docs/features.md as the product-facing feature map for README, website, screenshot, and demo work.
  • Add docs/memory-center.md and a Memory Graph/Dream spec to document the new memory workspace and safety model.
  • Refresh the README into a public product page with install paths, product surfaces, documentation map, development notes, security/community pointers, star history, and an expanded agent-context block.
  • Refresh the docs index, setup/configuration docs, customization docs, Plan Mode docs, Usage Insights docs, and package-sharing docs to point at implemented screenshots and provider-neutral examples.

Security and Release Integrity

  • Refresh release-gated dependencies and lockfile overrides for DOMPurify, Undici, form-data, esbuild, Slack Bolt, and OpenTelemetry so the OSV release scanner can pass without disabling supply-chain checks.

v0.1.14

GitHub

Package registry and setup release. It makes package installs target a selected package instead of an entire source, keeps source-level workflows explicit, and updates the release dependency gate.

Highlights

  • Make mendcode packages install <pack-id> install a selected package from the official or chosen registry catalog instead of applying a whole source by accident.
  • Keep source-level package application available as mendcode packages install-source <source-id> for existing local/team registry workflows.
  • Let the setup Package step skip packages, browse official packages, install a local package path, create/update a local package snapshot, or manage installed overlays.
  • Add registry coverage for multi-package catalogs, selected pack installation, source-level install compatibility, and runtime compatibility rejection.

Added

  • Add registry coverage for multi-package catalogs, selected pack installation, source-level install compatibility, and runtime compatibility rejection.

Fixed

  • Make mendcode packages install <pack-id> install a selected package from the official or chosen registry catalog instead of applying a whole source by accident.
  • Keep source-level package application available as mendcode packages install-source <source-id> for existing local/team registry workflows.

Changed

  • Let the setup Package step skip packages, browse official packages, install a local package path, create/update a local package snapshot, or manage installed overlays.

Security and Release Integrity

  • Update protobufjs to 7.6.3 so the release dependency gate can pass with the fixed version.

v0.1.13

GitHub

Installer, setup config, and runtime metadata release. It improves first-run install messaging, prevents MendCode-owned setup config from crashing runtime config loading, and hardens registry/release dependency metadata.

Highlights

  • Refresh the Bash and PowerShell installers with a clearer MendCode banner, numbered install phases, ASCII progress output, and more direct next-step guidance.
  • Make installer guidance honest after curl | bash: run the installed binary by absolute path immediately, or open/source a terminal before using mendcode from PATH.
  • Keep MendCode-owned setup config keys such as memory and package from crashing the runtime config loader after first-run setup creates .mendcode/mendcode.json.
  • Honor dynamic OPENCODE_CONFIG_CONTENT values at config-load time so SDK/runtime callers can inject per-instance config after process startup.

Fixed

  • Keep MendCode-owned setup config keys such as memory and package from crashing the runtime config loader after first-run setup creates .mendcode/mendcode.json.
  • Preserve explicit model changes made while a session is busy so the next prompt uses the newly selected model instead of falling back to the previous session model.
  • Point npm registry publishing at the public mendcode package name instead of the temporary mendcode-ai wrapper name.

Changed

  • Refresh the Bash and PowerShell installers with a clearer MendCode banner, numbered install phases, ASCII progress output, and more direct next-step guidance.
  • Make installer guidance honest after curl | bash: run the installed binary by absolute path immediately, or open/source a terminal before using mendcode from PATH.

Security and Release Integrity

  • Honor dynamic OPENCODE_CONFIG_CONTENT values at config-load time so SDK/runtime callers can inject per-instance config after process startup.
  • Update Hono to 4.12.25 so release dependency scanning passes without carrying known fixed advisories.

v0.1.12

GitHub

Windows installer and rich terminal rendering release. It adds Windows install paths, configurable assistant rendering modes, and better Markdown/Mermaid/table handling in chat and Plan Review.

Highlights

  • Show the upstream opencode provider in the TUI provider/model/setup flows while keeping it labeled as opencode Zen, and keep opencode-go labeled separately as opencode Go.
  • Add a Windows PowerShell installer and document macOS, Linux, Windows PowerShell, and Git Bash/WSL install paths in the README.
  • Add configurable assistant message rendering modes: Plain, Markdown, and Rich.
  • Render plan/chat Markdown more cleanly in Rich mode, including local Mermaid flowcharts, wide tables, checklists, callouts, connectors, and compact terminal-friendly diagrams.

Added

  • Show the upstream opencode provider in the TUI provider/model/setup flows while keeping it labeled as opencode Zen, and keep opencode-go labeled separately as opencode Go.
  • Add configurable assistant message rendering modes: Plain, Markdown, and Rich.
  • Render plan/chat Markdown more cleanly in Rich mode, including local Mermaid flowcharts, wide tables, checklists, callouts, connectors, and compact terminal-friendly diagrams.

Fixed

  • Keep plan review content anchored while scrolling so review modals stay readable during long plans.

Docs

  • Add a Windows PowerShell installer and document macOS, Linux, Windows PowerShell, and Git Bash/WSL install paths in the README.

v0.1.11

GitHub

Runtime reliability release for subagents, queued prompts, memory notifications, footer chrome, and mflow external-path handling.

Highlights

  • Preserve task tool metadata while a subagent is running, after tool-call transitions, and when a subagent fails before model resolution completes.
  • Queue a newly submitted prompt behind an active run so the second prompt is saved and answered in the next LLM call instead of being coalesced into the first run's result.
  • Keep existing concurrent loop() callers coalesced to the same active run while only explicit prompt submissions enqueue a follow-up turn.
  • Stop showing the empty memory-check toast when the extractor finds no durable candidates, and keep Home title changes silent while refreshing the visible Home surface.

Fixed

  • Preserve task tool metadata while a subagent is running, after tool-call transitions, and when a subagent fails before model resolution completes.
  • Queue a newly submitted prompt behind an active run so the second prompt is saved and answered in the next LLM call instead of being coalesced into the first run's result.
  • Keep existing concurrent loop() callers coalesced to the same active run while only explicit prompt submissions enqueue a follow-up turn.
  • Stop showing the empty memory-check toast when the extractor finds no durable candidates, and keep Home title changes silent while refreshing the visible Home surface.
  • Hide the prompt-bottom context meter unless promptStatus.context.visible is explicitly enabled, including old profiles that still list the context builtin.

v0.1.10

GitHub

Memory proposal and prompt-routing release. It adds approval-gated memory updates/removals, fixes slash-command false positives, improves Agent View behavior, and preserves pasted images.

Highlights

  • Let automatic memory learning create approval-gated add, update, and remove proposals, with targeted updates/deletes applied to existing memory entries instead of only appending new entries.
  • Recover durable repo-scoped memory rules when the extractor returns empty or wrapped JSON, without relying on hard-coded prompt examples.
  • Show a memory toast when extraction finishes with no pending proposal, including skipped or no-candidate reasons instead of silently disappearing after "Preparing memory proposal...".
  • Treat slash commands as commands only when they start the prompt, so normal messages that mention /setup, /stats, or other commands are sent as chat text instead of navigating away.

Added

  • Show a memory toast when extraction finishes with no pending proposal, including skipped or no-candidate reasons instead of silently disappearing after "Preparing memory proposal...".

Fixed

  • Let automatic memory learning create approval-gated add, update, and remove proposals, with targeted updates/deletes applied to existing memory entries instead of only appending new entries.
  • Treat slash commands as commands only when they start the prompt, so normal messages that mention /setup, /stats, or other commands are sent as chat text instead of navigating away.
  • Keep /setup and /stats as temporary pages from an active chat: pressing escape or finishing setup now returns to the originating session instead of dropping to New Chat.
  • Keep Agent View hover visual-only; click or arrow keys select the reply target, and esc clears that selected session before returning to normal prompt input.
  • Keep split home welcome branding neutral and prevent long ASCII title text from clipping into partial letters.

Docs

  • Recover durable repo-scoped memory rules when the extractor returns empty or wrapped JSON, without relying on hard-coded prompt examples.

v0.1.9

GitHub

Public CLI and compaction cleanup release. It simplifies mendcode --help, hides low-level internals from normal help, fixes overflow compaction replay behavior, and restores terminal modes on exit.

Highlights

  • Make mendcode --help workflow-first: open the terminal harness, run with an initial intent, use packages, mflow, worktrees, TSM, setup, and status.
  • Move low-level/debug commands out of normal help, including TUI profile internals, runtime adapter/upstream/export/config plumbing, and prompt/runtime internals.
  • Keep legacy aliases such as init, sync, package, and prompts callable with deprecation warnings instead of presenting them as product workflows.
  • Suggest close matches for typo-prone commands, including mendcode tui prewview -> mendcode tui preview.

Fixed

  • Keep legacy aliases such as init, sync, package, and prompts callable with deprecation warnings instead of presenting them as product workflows.
  • Stop replaying the latest real user message as a new visible user turn after overflow compaction; resume from the summary and a synthetic internal continue prompt instead.
  • Avoid the double-compaction path where auto resume immediately re-adds the same user request before continuing.
  • Release mouse tracking, bracketed paste, and raw input mode before suspending or exiting the TUI so the parent shell does not receive scroll/click escape sequences.

Changed

  • Make mendcode --help workflow-first: open the terminal harness, run with an initial intent, use packages, mflow, worktrees, TSM, setup, and status.
  • Move low-level/debug commands out of normal help, including TUI profile internals, runtime adapter/upstream/export/config plumbing, and prompt/runtime internals.
  • Suggest close matches for typo-prone commands, including mendcode tui prewview -> mendcode tui preview.

v0.1.8

GitHub

Usage Insights and installer reliability patch. It fixes loading-state UX, global stats cache usage, weather config persistence, and clean-home install version detection.

Highlights

  • Show a real Usage Insights loading state instead of zeroed metrics while cached stats are still loading.
  • Reuse the global TUI stats cache on the Usage Insights page without warming session messages during normal chat startup.
  • Keep the weather location in the global TUI config and simplify the stats shortcuts by removing the manual refresh action.
  • Fix the installer version check so a same-version global mendcode on PATH cannot falsely satisfy a clean $HOME/.mendcode/bin install.

Added

  • Show a real Usage Insights loading state instead of zeroed metrics while cached stats are still loading.

Fixed

  • Keep the weather location in the global TUI config and simplify the stats shortcuts by removing the manual refresh action.
  • Fix the installer version check so a same-version global mendcode on PATH cannot falsely satisfy a clean $HOME/.mendcode/bin install.

Changed

  • Reuse the global TUI stats cache on the Usage Insights page without warming session messages during normal chat startup.

v0.1.7

GitHub

Setup, memory extraction, compaction, paste visibility, and docs refresh release.

Highlights

  • Keep the setup screen stable while refreshing step state so changing setup pages no longer flashes back to a loading placeholder.
  • Widen the setup rail so longer step labels such as TUI Profile and Permissions do not collide with completion status.
  • Shorten and row-budget the memory extractor auth warning while preserving the actionable OAuth/client-id blocker.
  • Inject global memory at session start and after compaction, while keeping per-request project memory local to the model prompt instead of writing it into chat history.

Added

  • Inject global memory at session start and after compaction, while keeping per-request project memory local to the model prompt instead of writing it into chat history.

Fixed

  • Keep the setup screen stable while refreshing step state so changing setup pages no longer flashes back to a loading placeholder.
  • Widen the setup rail so longer step labels such as TUI Profile and Permissions do not collide with completion status.
  • Prevent the primary assistant from directly saving implicit preferences to memory; approval-gated proposals now come from the memory extractor unless the user explicitly asks to save memory immediately.
  • Avoid re-triggering assistant generation after clean auto-compaction while still resuming when compaction interrupted an active or incomplete assistant turn.
  • Keep long pasted content visible until it exceeds the configured character threshold, and let truncated user-message headers expand from the hidden-content hint without stealing normal message action clicks.

Changed

  • Shorten and row-budget the memory extractor auth warning while preserving the actionable OAuth/client-id blocker.
  • Run automatic memory extraction after normal assistant stops, show a dedicated memory activity state/mascot while proposals are generated, and give the extractor structured saved/pending memory context before it decides whether to propose.

Docs

  • Refresh the docs for the mendcode command surface, customization, setup/configuration, mflow, package sharing, plan mode, usage insights, TSM/worktrees, and wiki navigation.

v0.1.6

GitHub

Usage Insights and prompt-memory release. It adds the TUI usage dashboard and request-mode memory injection/search behavior.

Highlights

  • Add a TUI Usage Insights dashboard with daily token activity, global/project stats, response-time metrics, top tools/agents/models, cached stats loading, and optional Open-Meteo weather.
  • Inject saved global memories into normal runtime requests as transient system context instead of only after compaction.
  • Keep project memories in the runtime prompt with their own request cap so repo-local context is available without persisting injected memory into chat history.
  • Make mendcode memory search and mendcode memory preview show request-mode retrieval by default, with an explicit --mode selector for compaction/manual checks.

Added

  • Add a TUI Usage Insights dashboard with daily token activity, global/project stats, response-time metrics, top tools/agents/models, cached stats loading, and optional Open-Meteo weather.
  • Inject saved global memories into normal runtime requests as transient system context instead of only after compaction.

Fixed

  • Keep project memories in the runtime prompt with their own request cap so repo-local context is available without persisting injected memory into chat history.

Changed

  • Make mendcode memory search and mendcode memory preview show request-mode retrieval by default, with an explicit --mode selector for compaction/manual checks.

v0.1.5

GitHub

OpenCode MCP/runtime backport release with provider isolation, subagent identity preservation, and shell busy-state race fixes.

Highlights

  • Backport MCP runtime updates: catalog pagination, capability-aware prompt/resource listing, tolerant tool schema discovery, abort signal forwarding, OAuth callbackPort/scope, and manual OAuth header preservation.
  • Isolate provider model plugin hooks from internal provider state so plugin mutations cannot rename providers or zero model pricing globally.
  • Preserve delegated subagent identity on task child sessions.
  • Avoid a shell-cancel race by cancelling an existing session runner during short busy-state transitions instead of marking the session idle too early.

Fixed

  • Preserve delegated subagent identity on task child sessions.
  • Avoid a shell-cancel race by cancelling an existing session runner during short busy-state transitions instead of marking the session idle too early.

Changed

  • Backport MCP runtime updates: catalog pagination, capability-aware prompt/resource listing, tolerant tool schema discovery, abort signal forwarding, OAuth callbackPort/scope, and manual OAuth header preservation.
  • Isolate provider model plugin hooks from internal provider state so plugin mutations cannot rename providers or zero model pricing globally.

v0.1.4

GitHub

Compaction and prompt-mode correctness patch. It stops completed turns from being re-triggered by normal compaction and fixes Tab mode label state.

Highlights

  • Stop normal automatic and manual compaction from re-triggering the assistant after a completed turn.
  • Keep overflow compaction recovery for requests that really hit provider context limits.
  • Make Tab mode cycling update the active prompt mode instead of leaving the mode label stuck while model state changes.

Fixed

  • Stop normal automatic and manual compaction from re-triggering the assistant after a completed turn.
  • Keep overflow compaction recovery for requests that really hit provider context limits.
  • Make Tab mode cycling update the active prompt mode instead of leaving the mode label stuck while model state changes.

v0.1.3

GitHub

Agent View, local version display, prompt-context cleanup, and selected OpenCode terminal/SDK backports.

Highlights

  • Show richer Agent View welcome timestamps, including Today plus contextual day, month, and year information.
  • Show the local package version in the welcome screen instead of local when running from source.
  • Clarify prompt-context commands in the TUI by renaming the persistent selector and hiding the low-value cycle command from Ctrl+P.
  • Backport OpenCode terminal and SDK fixes for structured SDK errors, Basic Auth defaults, local MCP cwd, and MCP prompt/resource timeouts.

Added

  • Show richer Agent View welcome timestamps, including Today plus contextual day, month, and year information.
  • Show the local package version in the welcome screen instead of local when running from source.

Changed

  • Clarify prompt-context commands in the TUI by renaming the persistent selector and hiding the low-value cycle command from Ctrl+P.
  • Backport OpenCode terminal and SDK fixes for structured SDK errors, Basic Auth defaults, local MCP cwd, and MCP prompt/resource timeouts.
  • Stamp release binaries with the requested MENDCODE_VERSION/OPENCODE_VERSION and include release workflow changes in generated notes.

v0.1.2

GitHub

CLI naming, update health, and metadata alignment release for the public mendcode command.

Highlights

  • Publish MendCode under the mendcode command name and remove the public mend package alias.
  • Normalize source and package metadata to 0.1.2 after the public v0.1.1 release.
  • Show the MendCode runtime version in welcome, CLI, debug info, health responses, and Zed extension metadata.
  • Keep update checks on MendCode/MendCode GitHub releases and skip autoupdate for local/source builds.

Added

  • Publish MendCode under the mendcode command name and remove the public mend package alias.
  • Show the MendCode runtime version in welcome, CLI, debug info, health responses, and Zed extension metadata.

Fixed

  • Keep update checks on MendCode/MendCode GitHub releases and skip autoupdate for local/source builds.
  • Preserve offline startup by treating update lookup failures as non-blocking.

Changed

  • Normalize source and package metadata to 0.1.2 after the public v0.1.1 release.
  • Replace user-facing mend and mend-runtime command hints with mendcode.

v0.1.1

GitHub

Small public release with TUI context inspection and release-infrastructure hardening.

Highlights

  • Add a TUI Context usage dialog for inspecting current chat context usage, token breakdown, cache read/write, reasoning, free context, turns, tool calls, and compaction count.
  • Add /context as a native TUI slash command so the context usage dialog can open directly from the prompt.
  • Clarify agent mode controls in the TUI.
  • Reserve native UI slash command names so synced MCP/skill commands cannot shadow built-in UI commands such as /context.

Added

  • Add a TUI Context usage dialog for inspecting current chat context usage, token breakdown, cache read/write, reasoning, free context, turns, tool calls, and compaction count.
  • Add /context as a native TUI slash command so the context usage dialog can open directly from the prompt.

Changed

  • Clarify agent mode controls in the TUI.
  • Reserve native UI slash command names so synced MCP/skill commands cannot shadow built-in UI commands such as /context.

Security and Release Integrity

  • Update pinned GitHub Actions to Node 24-compatible commits while keeping SHA pins.
  • Keep release builds behind supply-chain preflight, OSV lockfile gate, frozen lockfile install, disabled dependency scripts, SBOM generation, artifact attestations, and checksum generation.