MicroClaw: oudere versies downloaden

MicroClaw Code 0.31: private project memory, Claude Code plugins and idle
compaction (docs/roadmap/mccode-competitive-plan-2026-11-cn.md, N5–N7).
Added
- Private project memory. Besides the shared .microclaw/memory.md
(in the repository), trusted projects now have a private memory in
~/.mccode/projects/<repo>-<hash>/memory.md for your preferences and this
machine's setup. Both are in the prompt of every turn; the agent is told
which kind of fact goes where (a stated preference never goes to the
shared file), and writing the private file does not ask for the
outside-the-project approval. Worktrees share the main checkout's private
memory. /memory opens both files in $EDITOR (shared, private,
show). mccode import claude adds Claude Code's auto memory of the same
repository, mccode import codex the items of Codex's memories.
- Plugins in the Claude Code layout. mccode plugin add <owner/repo | git URL | path> [--plugin <name>] [--project], list, update,
enable, disable and remove. A plugin's commands (/name and
/plugin:name), agents (for explore and sessions_spawn), skills,
hooks and MCP servers work as in Claude Code, with ${CLAUDE_PLUGIN_ROOT}
resolved; marketplace repositories are followed to the plugin's own
repository. Installing lists every hook and MCP command and asks first
(--yes without a terminal); updating asks again when new commands
appear. Unsupported parts (output styles, LSP servers, …) are listed as
skipped. User plugins are copied to ~/.mccode/plugins/, project plugins
to .microclaw/plugins/ and load only in trusted projects. /plugins
shows what is in use. Engine: extra_agents_dirs, extra_skills_dirs,
extra_mcp_configs; SDK: MicroClawBuilder::plugin_sources.
- Idle compaction. The terminal UI compacts a large session (at least
half of the compaction budget) that has been idle for 90% of the
provider's prompt cache lifetime, while the cache is still warm, so the
next prompt does not write the whole conversation to the cache again.
Anthropic only (the engine reports prompt_cache_ttl in
ContextUsage); tui.idle_compact: false turns it off. pre_compact
hooks see trigger auto.
- MicroClaw Code desktop app (preview, macOS). A GPUI app in
apps/microclaw-code-desktop with onboarding, projects, sessions and a
streaming transcript over the same ~/.mccode home and Agent Engine as
the terminal UI. Build it from source with scripts/bundle.sh; it is
not signed or shipped as a download yet.
Fixed
- glob no longer scans the disk for hours: its fallback for hidden,
ignored and absolute paths walks from the pattern's literal prefix and
stops after 50,000 entries (saying so), off the async workers.
- Plan mode and read-only profiles offer activate_skill, which only reads
a skill's instructions.
- mccode -p "/unknown" reports the unknown command instead of sending an
empty turn.
- The system prompt explains the <context_budget> line, which models
otherwise read as text injected into tool output.

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.30.1...v0.30.2

MicroClaw Code closes four everyday gaps found by comparing it with Claude
Code, Codex CLI and OpenCode
(docs/roadmap/mccode-competitive-plan-2026-11-cn.md, N1–N4).
Added
- Ctrl+B moves a running shell command to the background. The agent's
bash call returns at once with the process id and the agent carries
on; it reads the output later with bash_output and stops it with
bash_kill. /jobs (terminal and Zed) lists background commands and
sub-agent runs, /jobs <id> shows the latest output and /jobs kill <id>
stops one, also while a turn runs. Foreground commands in a workspace now
run through the process registry for this (stdout and stderr arrive
interleaved, as in a terminal); a cancelled run still stops its command.
SDK: MicroClaw::background_running_commands, background_jobs,
background_job_output, kill_background_job.
- /rewind also rewinds the conversation. Going back to before a turn
restores its files and cuts the conversation back to where it stood when
the turn started, so the agent no longer remembers work that was undone.
The timeline (Esc Esc) takes Enter for both, f for files only, c for
the conversation only and e to just edit the prompt; /rewind <id> files|chat|both and Zed's /rewind N [files|chat] do the same. After a
files-only rewind, /undo or a conversation-only rewind, the next prompt
tells the agent that the files and the conversation may disagree. The
whole conversation is archived before it is cut, and a conversation that
was compacted since the turn is left as it is. SDK:
MicroClaw::session_mark and rewind_session.
- Sub-agents get their own model, and project agents can run as
sub-agents. model.roles.subagent (engine aux_models.subagent, SDK
MicroClawBuilder::subagent_model) now sets the model of sessions_spawn
runs and explore. sessions_spawn takes agent: <name> to run a task
as a project agent from .microclaw/agents/<name>.md with its own
instructions and model. A new tools: front-matter field (tool names
or read, write, edit, shell, search, web) narrows the tools
such an agent may use, for explore as well; it can never widen them.
- The approval reviewer can approve. permissions.reviewer: approve
lets the reviewer model approve, in ask mode, the high-risk calls it
rates SAFE (shell commands, writes outside the project, web fetches, MCP
tools); everything else still asks you, destructive commands and ask
rules always do, and a failed review asks too. The turn's receipt lists
what the reviewer approved and why. advise is the former
--review-commands behaviour (the flag still selects it), and the
reviewer now runs on model.roles.review.
Changed
- The side panel moves from Ctrl+B to Ctrl+] (Ctrl+5 in terminals without
the kitty keyboard protocol); rebind app.sidebar / app.background in
tui.keymap to change either.
- The approval reviewer's prompt says that a SAFE verdict may let a call run
without asking a person, so it keeps SAFE for clearly harmless calls.

Fixed
- openai-codex no longer stops working when its OAuth token expires.
The token from ~/.codex/auth.json was read and refreshed only when the
provider was created at startup, so a long-running Server or mccode
session kept sending an expired token and every request failed with
HTTP 401 until a restart. The auth file is now read before every request,
the token is refreshed (and written back) five minutes before it expires,
and a 401 triggers one refresh and retry. Concurrent refreshes are
serialised, so a rotated refresh token is never used twice. The startup
refresh also used a blocking HTTP client inside the async runtime, which
could panic when the token had already expired.

/goal works everywhere MicroClaw Code runs: the terminal, Zed and
mccode -p. Zed also gains the common terminal commands.
Added
- /goal in Zed and in mccode -p, sharing one loop with the terminal
UI (microclaw-code-core::goal). --until <command> makes a command the
judge: the goal is reached only when it succeeds, it runs after every
round, and its failing output goes into the next round. --rounds N
sets the limit (default 8). In Zed one prompt runs every round and the
stop button ends the goal; mccode -p "/goal …" exits 1 when the goal
was not reached, and --json adds a goal object to the result.
- MicroClaw Code in Zed gains /btw, /check, /diff, /rewind [N],
/status, /export and /init, so the common terminal commands work
there too. Trust stays a terminal decision (mccode trust).
Changed
- /goal is stricter: GOAL COMPLETE counts only as the reply's last
line, a reply ending in GOAL BLOCKED: <reason> stops the goal and says
why, and two rounds in a row that change no files stop it as stalled
instead of using up the remaining rounds.

The Matrix channel is removed, and security advisories are fixed. Matrix
went because its SDK pinned a vulnerable SQLite for every product built
from this repository, and nothing showed that anyone used it.
Removed
- The Matrix channel (channel-matrix), the full Cargo feature and
the microclaw-full release archives, Homebrew formula, install.sh --full and install.ps1 -Full variant. The full variant differed from
the default build only by Matrix. A channels.matrix block in an existing
config is ignored with a warning. In the Homebrew tap, microclaw-full
now renames to microclaw, so brew update moves existing installs over.
--full and -Full are still accepted by the install scripts and only
print a notice.
Security
- rusqlite 0.41 (from 0.40), which bundles SQLite 3.54.1 through
libsqlite3-sys 0.39.1, for RUSTSEC-2026-0459 (CVE-2026-119375): SQLite
before 3.54.0 could return uninitialized page-cache memory for a database
opened with a crafted -wal file. Every product bundles SQLite, so all
were affected; the Matrix SDK's pin on rusqlite 0.40 blocked the
upgrade until the channel was removed. Six cargo deny exceptions that
only applied to Matrix's dependencies are gone too.
- rmcp 3.9 (from 1.8) for RUSTSEC-2026-0462: the MCP Streamable HTTP
client read response bodies with no size limit, so a malicious or
compromised MCP server could exhaust memory, even during initialize.
Successful JSON bodies are now capped at 16 MiB and error bodies at
64 KiB. Affects the Server, MicroClaw Code and the SDK wherever HTTP MCP
servers are configured.
- openssl 0.10.82 (from 0.10.76) for RUSTSEC-2026-0463, an out-of-bounds
write in aes::unwrap_key; it reaches the build through native-tls.

Fixes found by testing MicroClaw Code 0.27.0 in Zed and in the terminal
with Anthropic's claude-opus-5, including image prompts.
Fixed
- Anthropic usage left out prompt-cache tokens (all products).
Anthropic's input_tokens excludes cache reads and writes, so with
prompt caching a request counted only a few input tokens. Token counts
and costs were too low, /context and the context meter showed 0%, and
mid-turn context-pressure compaction measured the wrong number. Cache
reads and writes now count as input, streamed and not, as they already
did for OpenAI-compatible providers. A final message_delta that carries
only output_tokens no longer drops the input count.
- MicroClaw Code's model pickers (Zed's model selector and the TUI
/model list) offered preset family names such as claude-sonnet,
claude-opus and claude-1m. These are not model names, and choosing
one failed with not_found_error. The pickers now offer only presets
that are real model names (ModelPreset::model_name).
- In Zed, a shell command that needed approval kept the title
"Allow bash?" after the answer. The permission request now keeps the
call's own title, the command; only an approval with no visible call is
titled "Allow ?".
- A failed run in mccode acp, such as an unknown model or a provider
error, was reported as a thought or an info notice, so the turn looked
empty. It is now part of the reply ("Error: …").
- Tool titles in mccode acp are Markdown code spans, so a path like
calc/__init__.py no longer renders as bold "init".

MicroClaw Code becomes a first-class agent in Zed (mccode acp), and the
website and docs now recommend Zed as the main way to use it, with the
terminal interface as the alternative.
Added
- MicroClaw Code in Zed (mccode acp) now speaks ACP protocol v1 with schema
1.15: mode and model selectors (session config options), slash commands
(/review, /compact, /undo, /cost, /context, /help and custom
commands), the context and cost indicator, real file diffs with old and new
text, absolute paths so Zed can follow the agent, project checks as their
own tool cards, thread titles and history (session/list, resume, close),
notices and compaction updates, and a sign-in prompt that runs
mccode setup in a terminal when no model is configured.
- scripts/acp_registry_entry.py writes the ACP Registry entry for a release
(packaging/acp-registry/).
Fixed
- mccode acp and mccode mcp-server use model profiles from
settings.yaml; before, they required a configuration file and stopped at
startup without one. They also apply permissions.default_mode,
disabled_modes and the project's compaction settings.
- mccode config show and config get no longer print API keys; they show
sk-…abcd.
- mccode acp lists every project's sessions when session/list has no
cwd, so Zed's "Import Threads" finds them; before, it returned none.
- Edit diffs in Zed no longer come out empty when the edit had already been
written by the time its start event was handled: the text before the edit
is now rebuilt from the edit's own diff.
- A prompt that is only an attached file (Zed's "+" menu) names the thread
after the file instead of its absolute path.
- /config and /config get in the terminal UI mask API keys like
mccode config show does; before, they printed them in full.
- Docs: mccode export takes --session before the subcommand, --trust
saves the trust decision, the sandbox also hides ~/.mccode,
MCCODE_LOCAL_SANDBOX accepts landlock or none, "always allow this
command" rules live in ~/.mccode/allow-rules.json, chat notifications
need their event in notify.events, and admonitions on the website render
again (Docusaurus v4 syntax).
Changed
- Engine setup for every Code front end lives in
microclaw_code_core::launch.
- docs/schema/code-settings.v1.json is written with sorted keys, so it is
the same whichever crates share the build.
- The website, docs and READMEs recommend Zed as the main way to use
MicroClaw Code (Agent Panel through mccode acp), with the terminal
interface as the alternative; the editors page is now "Use with Zed" and
comes first in the Code docs.
- The Code docs, READMEs and the website's Code card now help you get Zed
first: a download link, the Homebrew, install-script and winget commands
with system requirements, the "Add Custom Agent" route in Zed's agent
settings, and a note that the Zed integration needs mccode 0.27.0 or later.

MicroClaw Code becomes a fully separate product from the MicroClaw Server.
After upgrading, run mccode setup (or add a model profile to
~/.mccode/settings.yaml); an old ~/.microclaw/code is not read.
Changed
- MicroClaw Code is fully separate from the MicroClaw Server. Its home
moves from ~/.microclaw/code to ~/.mccode (nothing is migrated: move the
old directory by hand to keep settings and sessions). Code no longer
borrows model credentials from MICROCLAW_CONFIG or
~/.microclaw/microclaw.config.yaml: it uses --config, MCCODE_CONFIG,
~/.mccode/config.yaml or a settings profile. Its variables are
MCCODE_HOME, MCCODE_CONFIG, MCCODE_SANDBOX, MCCODE_MANAGED,
MCCODE_LOCAL_SANDBOX, MCCODE_PATH_ALLOWLIST and MCCODE_REPO (the old
MICROCLAW_CODE_* names are no longer read), and Server variables such as
MICROCLAW_CONFIG, MICROCLAW_SKILLS_DIR and MICROCLAW_LOCAL_SANDBOX
no longer affect it. Managed settings move to /etc/mccode/managed.yaml
(macOS /Library/Application Support/mccode/managed.yaml, Windows
%ProgramData%\mccode\managed.yaml). The OS sandbox hides ~/.mccode
from shell commands, like ~/.microclaw.
- Documentation and the mccode config init template keep role models on the
session's profile, and say that model.roles.subagent is not used yet.

The MicroClaw Code interaction and configuration release. The planned 0.24
(TUI), 0.25 (settings and permissions) and 0.26 (innovations) ship together;
see docs/roadmap/mccode-ux-config-plan-2026-11-cn.md.
Added
- TUI look: theme tokens with truecolor, 256- and 16-colour fallback;
11 built-in themes, user YAML themes and OpenCode theme files; Markdown
replies with syntax highlighting; diffs with line numbers, word-level
highlights and a side-by-side layout on wide terminals; consecutive reads
and searches grouped into one "Explored" row; a side panel for the plan,
checks, changed files and running sub-agents; a per-item render cache; an
inline mode that uses the terminal's own scrollback (tui.screen: inline,
falling back to full screen when the terminal does not report the cursor).
- TUI interaction: Enter queues a message while the agent works and
Ctrl+S steers the running turn; fuzzy / and @ pop-ups and a Ctrl+P
palette; per-project prompt history with Ctrl+R search; Ctrl+G edits in
$EDITOR; ! runs a shell command yourself; large pastes become a chip;
Ctrl+V attaches a clipboard image; Ctrl+T opens the transcript with search;
Esc Esc rewinds the files to an earlier turn; model, theme (live preview)
and session pickers; mouse scrolling and click-to-expand; desktop
notifications (OSC 9/777) when the terminal is unfocused; rebindable keys
(tui.keymap, /keys).
- Per-turn receipts (files, checks, sandbox and network use, tokens, cost
when the price is known, checkpoint), hunk-level review on F2 (revert a
hunk, or turn comments into a prompt), /goal (keep going until a
condition holds and the checks pass), /btw (a side question in a copy
of the session) and /configure (describe a settings change; the model
proposes it and you confirm).
- Layered settings for MicroClaw Code: user, project, local and managed
YAML layers with a JSON schema (docs/schema/code-settings.v1.json) and
per-key provenance; mccode config show|get|set|unset|paths|doctor|init|schema|migrate|edit;
mccode import claude|codex|opencode; provider profiles that can replace
the configuration file; per-role models (plan, review, summary).
- Permission rules (allow / ask / deny for bash, read, edit,
net, mcp, agent, tool; Claude Code spellings accepted) and hooks
compatible with the Claude Code protocol (ten events; exit code 2 blocks;
permissionDecision, updatedInput, additionalContext). Both are
enforced through a new engine ToolGate (SDK MicroClawBuilder::tool_gate),
so sub-agents go through them too. ACP and MCP-server modes load the same
settings.
- Approvals and notifications in chat: with a paired Server, MicroClaw
Code sends approval cards (buttons on Telegram, Discord and Slack; numbered
options on Feishu) and "done" notifications while you are away
(notify.chat); Server endpoints /api/code/notify and
/api/code/approvals.
- SDK: MicroClawBuilder::summary_model; the ToolGate types are re-exported.
Fixed
Found by testing this release with DeepSeek (deepseek-flash,
deepseek-v4-pro):
- The shell tool ran commands in the login shell ($SHELL); zsh aborts on an
unmatched glob such as grep --include=*.rs. Commands now run in bash when
it is installed (Server and Code).
- bash_kill stopped only the shell, so a child (a dev server under
npm run dev) kept running and the process stayed "running". Background
commands now run in their own process group and are stopped as a group.
- Role models sent profile/model to the provider as the model name; an
ask rule could be answered by auto mode; -p 1 after a pause looped when
a prompt hook added context to the answer.
Changed
- Writing .microclaw/settings.local.yaml adds it to the repository's
.git/info/exclude.
- scripts/generate_docs_artifacts.mjs leaves #[serde(skip)] config fields
out of the generated defaults.
Security
- vodozemac 0.11.2 (Matrix end-to-end encryption), which restores secret
zeroization (RUSTSEC-2026-0445, published on release day).
Known gaps
Ctrl+B does not move a running command to the background (it toggl

Fixes found by running MicroClaw Code against real models (DeepSeek
deepseek-flash and deepseek-v4-pro) on the eval cases and by hand.
Fixed
- Approvals resumed later were not understood: answering a paused
approval with 1/2 or 3 in a later turn (mccode --session <id> -p 1,
the MCP run tool, and the Telegram / Discord / Slack / web option cards)
reached the model as a bare digit, which some models did not read as an
approval. The reply is now rewritten into an explicit instruction to retry
(or not run) the paused call.
- Compaction never ran with small context windows: the verbatim tail kept
by a Code compaction (24k tokens) did not shrink with the target, so a
session on a small window (a local model, an eval case) could not be
compacted at all and ran into a context overflow. The tail is now capped at
half of the compaction target.
- DeepSeek V4 windows: deepseek-v4-pro and deepseek-flash
(deepseek-v4-flash) are counted as 1M-token models instead of 128k.
- Language-server diagnostics:
- pyright never published diagnostics because MicroClaw Code did not send
workspace/didChangeConfiguration;
- a server on PATH that cannot start (rustup's rust-analyzer proxy
without the component) was shown as a passing check; it is now reported
once with an install hint and skipped;
- a server that sends nothing in time is reported as not checked;
- when diagnostics were the only check, a fix round that made them pass
still ended -p with exit code 1.
- @image mentions followed by punctuation (@shot.png?, @shot.png。)
were not attached.
- mccode export failed without a terminal ("the interactive interface
needs a terminal").
- --worktree summary suggested git merge while the changes were still
uncommitted in the worktree; it now says to commit them first.
- Repositories without a .gitignore: review lists at most 100
untracked files (the rest counted per directory), and the per-turn change
fingerprint no longer reads every large untracked file.
Changed
- mccode eval: min_compactions in case.yaml fails a case that did not
compact; the long-session cases use windows above the fixed prompt size so
they really compact, and the compaction count no longer races the run.
Security
- hyper 1.12.0, which drops the unmaintained pin-utils
(RUSTSEC-2026-0441, published on release day).
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.23.0...v0.23.1

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.11.0...v0.11.1

What's Changed
- test(web): Playwright browser smoke test for the embedded Web UI by @everettjf in https://github.com/microclaw/microclaw/pull/522
- web: move to @assistant-ui/react 0.15 and drop the tap pin by @everettjf in https://github.com/microclaw/microclaw/pull/523
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.10.1...v0.11.0

Fixed
- Web UI: the first keystroke in the message composer threw
Maximum update depth exceeded and froze the page (#517). The 2026-10-05
dependency bump let the transitive @assistant-ui/tap float to 0.9.20,
whose getSnapshot loop guard fires against the unfixed
@assistant-ui/core 0.2.23 that the 0.14 line is frozen on; v0.8.0,
v0.9.0 and v0.10.0 all shipped that pair. web/package.json now pins
@assistant-ui/tap to 0.9.12 through overrides, and Dependabot holds it
there until the UI moves to @assistant-ui/react 0.15.
Security
- Targeted dependency updates for advisories published after 0.10.0:
ctutils 0.4.5 (RUSTSEC-2026-0385), rustls-webpki 0.103.16
(RUSTSEC-2026-0382), tokio 1.54.1 (RUSTSEC-2026-0386), webpki-root-certs
1.0.11 (RUSTSEC-2026-0383) and webpki-roots 1.0.11 (RUSTSEC-2026-0381).

What's Changed
- nix: follow the nixpkgs automation/AI policy in update-nixpkgs.sh by @everettjf in https://github.com/microclaw/microclaw/pull/516
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.9.0...v0.10.0

What's Changed
- chore(deps): upgrade matrix-sdk to 0.19 (#510) by @everettjf in https://github.com/microclaw/microclaw/pull/513
- build(deps): bump @assistant-ui/react-markdown from 0.14.4 to 0.14.7 in /web in the web-production-non-major group across 1 directory by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/514
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.8.0...v0.9.0

What's Changed
- nix: clean up nixpkgs package template (drop unused inputs, run tests, version check, update script) by @everettjf in https://github.com/microclaw/microclaw/pull/494
- nix: enable __structuredAttrs in nixpkgs template by @everettjf in https://github.com/microclaw/microclaw/pull/495
- build(deps-dev): bump the web-development-non-major group in /web with 4 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/497
- build(deps): bump the cargo-patch group with 4 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/498
- fix(tools): block file-tool writes to governance files (#501) by @everettjf in https://github.com/microclaw/microclaw/pull/502
- fix(telegram): forward partial-quote context to the agent by @nilo85 in https://github.com/microclaw/microclaw/pull/506
- fix(memory): refresh instead of duplicating write_memory rows (#505) by @everettjf in https://github.com/microclaw/microclaw/pull/507
- chore(deps): hold @assistant-ui/react <0.15 and Zed git crates in dependabot by @everettjf in https://github.com/microclaw/microclaw/pull/503
- build(deps): bump the cargo-minor group across 1 directory with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/499
- ci(audit): ignore unreachable matrix-sdk advisories until the 0.19 upgrade by @everettjf in https://github.com/microclaw/microclaw/pull/509
- chore(deps): hold @assistant-ui/react-markdown below 0.14.8 by @everettjf in https://github.com/microclaw/microclaw/pull/511
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.7.0...v0.8.0

What's Changed
- nix: make update-nixpkgs.sh able to refresh the open nixpkgs init PR by @everettjf in https://github.com/microclaw/microclaw/pull/493
- fix(engine): keep the cached system prompt stable across turns by @nilo85 in https://github.com/microclaw/microclaw/pull/492
New Contributors
- @nilo85 made their first contribution in https://github.com/microclaw/microclaw/pull/492
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.6.1...v0.7.0

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.6.0...v0.6.1

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.0...v0.6.0

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.5.4...v0.5.5

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.5.3...v0.5.4

What's Changed
- fix: harden v0.4.0 approval/alerts/trust-tier flows + polish by @everettjf in https://github.com/microclaw/microclaw/pull/480
- Grok Build / grok-telegram-bot adoption plan: implement all P0–P3 items by @everettjf in https://github.com/microclaw/microclaw/pull/481
- v0.5.0: monolith decomposition sweep, vite 8/React 19, build fixes, docs-site dark theme by @everettjf in https://github.com/microclaw/microclaw/pull/482
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.4.0...v0.5.0

What's Changed
- build(deps-dev): bump the web-development-non-major group in /web with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/476
- Revert "build(deps-dev): bump the web-development-non-major group in /web with 2 updates" by @everettjf in https://github.com/microclaw/microclaw/pull/477
- v0.4.0 workstreams: competitive intel, plan, and full implementation by @everettjf in https://github.com/microclaw/microclaw/pull/478
- release: prepare v0.4.0 (web approval buttons, benchmark script portability) by @everettjf in https://github.com/microclaw/microclaw/pull/479
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.5...v0.4.0

What's Changed
- build(deps): bump the cargo-non-major group with 17 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/472
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.4...v0.3.5

What's Changed
- Add governed Learning Foundry pipeline by @everettjf in https://github.com/microclaw/microclaw/pull/470
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.3...v0.3.4

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.2...v0.3.3

What's Changed
- Durable Coworker and Secure Runtime by @everettjf in https://github.com/microclaw/microclaw/pull/464
- Add verified long-horizon learning by @everettjf in https://github.com/microclaw/microclaw/pull/465
- feat: add task-aware evaluated skills and recovery by @everettjf in https://github.com/microclaw/microclaw/pull/466
- feat: add comparative reflection and governed skill evolution by @everettjf in https://github.com/microclaw/microclaw/pull/468
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.1...v0.3.2

What's Changed
- ci: add tag-release workflow for cutting release tags via dispatch by @everettjf in https://github.com/microclaw/microclaw/pull/462
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.3.0-rc.1...v0.3.1

What's Changed
- feat(telegram): register command menu via setMyCommands by @everettjf in https://github.com/microclaw/microclaw/pull/436
- chore: rename microclaw.ai domain to microclaw.org by @everettjf in https://github.com/microclaw/microclaw/pull/438
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.2.5...v0.2.6

What's Changed
- feat(gateway): warn when systemd lingering is disabled on install by @everettjf in https://github.com/microclaw/microclaw/pull/434
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.2.3...v0.2.5

What's Changed
- feat(config): per-task auxiliary models, wired to context compaction by @everettjf in https://github.com/microclaw/microclaw/pull/403
- feat(config): route memory reflector through aux_models by @everettjf in https://github.com/microclaw/microclaw/pull/406
- feat: add microclaw eval trajectory evaluation gate by @everettjf in https://github.com/microclaw/microclaw/pull/408
- ci: enforce the microclaw eval trajectory gate by @everettjf in https://github.com/microclaw/microclaw/pull/410
- feat: sleep-time memory consolidation (deterministic, off by default) by @everettjf in https://github.com/microclaw/microclaw/pull/412
- feat(eval): detect stuck loops and consecutive tool-error streaks by @everettjf in https://github.com/microclaw/microclaw/pull/414
- feat(skills): deterministic skill audit curation command by @everettjf in https://github.com/microclaw/microclaw/pull/416
- feat(config): finish aux_models slots — title and vision by @everettjf in https://github.com/microclaw/microclaw/pull/417
- feat(audit): tamper-evident hash-chained audit log + verify CLI by @everettjf in https://github.com/microclaw/microclaw/pull/418
- ci: fix security-audit and docs-check pipeline failures by @everettjf in https://github.com/microclaw/microclaw/pull/419
- feat(chat): add in-chat /help command for command discovery by @everettjf in https://github.com/microclaw/microclaw/pull/420
- feat(config): friendlier parse errors with did-you-mean suggestions by @everettjf in https://github.com/microclaw/microclaw/pull/421
- feat(doctor): surface misconfigured channels with actionable messages by @everettjf in https://github.com/microclaw/microclaw/pull/422
- feat(channels): actionable auth-failure messages on startup by @everettjf in https://github.com/microclaw/microclaw/pull/423
- feat(doctor): live LLM credential check via doctor --online by @everettjf in https://github.com/microclaw/microclaw/pull/424
- feat(setup): refuse to save a config with no channel enabled by @everettjf in https://github.com/microclaw/microclaw/pull/425
- feat(doctor): check data/working dir writability by @everettjf in https://github.com/microclaw/microclaw/pull/426
- feat(cli): add Examples section to --help by @everettjf in https://github.com/microclaw/microclaw/pull/427
- feat(setup): label optional fields, not just required ones by @everettjf in https://github.com/microclaw/microclaw/pull/428
- feat(cli): post-setup next steps + per-subcommand --help examples by @everettjf in https://github.com/microclaw/microclaw/pull/429
- build(deps): bump the cargo-non-major group with 11 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/397
- Surface clawhub install warnings + guard eval negative fixtures by @everettjf in https://github.com/microclaw/microclaw/pull/430
- fix: handle orphaned tool_calls in DeepSeek reasoning bridge by @raaaas in https://github.com/microclaw/microclaw/pull/432
- feat(tools): opt-in RTK command rewriting for the bash tool by @everettjf in https://github.com/microclaw/microclaw/pull/433
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.2.2...v0.2.3

What's Changed
- fix(weixin): split long outbound messages to avoid truncation by @everettjf in https://github.com/microclaw/microclaw/pull/399
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.2.1...v0.2.2

What's Changed
- feat: add propagation-trace built-in skill by @everettjf in https://github.com/microclaw/microclaw/pull/384
- Make the bot more human + a capable concurrent specialist team, with 30 factory-ready skills by @everettjf in https://github.com/microclaw/microclaw/pull/391
- Humanlike follow-ups: progress-reports toggle, relationship familiarity, task ETA by @everettjf in https://github.com/microclaw/microclaw/pull/392
- Bounded versions of research-hard traits: humor timing, per-user growth, group interjection by @everettjf in https://github.com/microclaw/microclaw/pull/393
- Specialist-to-specialist collaboration via consult_specialist by @everettjf in https://github.com/microclaw/microclaw/pull/394
- Graph-augmented memory recall over the temporal knowledge graph by @everettjf in https://github.com/microclaw/microclaw/pull/395
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.59...v0.2.1

What's Changed
- feat: Add support for Telegram reply chains by @indrora in https://github.com/microclaw/microclaw/pull/383
New Contributors
- @indrora made their first contribution in https://github.com/microclaw/microclaw/pull/383
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.58...v0.1.59

What's Changed
- fix: UTF-8 safe string slicing in memory_backend.rs and web.rs by @loukurt-byte in https://github.com/microclaw/microclaw/pull/381
New Contributors
- @loukurt-byte made their first contribution in https://github.com/microclaw/microclaw/pull/381
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.57...v0.1.58

What's Changed
- feat: real-time mid-turn injection ack on non-web channels + close #307 umbrella by @everettjf in https://github.com/microclaw/microclaw/pull/345
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.56...v0.1.57

What's Changed
- build(deps-dev): bump the web-non-major group in /web with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/340
- fix(install): rename $pid param in install.ps1 to avoid PowerShell read-only var by @everettjf in https://github.com/microclaw/microclaw/pull/344
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.55...v0.1.56

What's Changed
- feat: port 6 hermes-agent features (prompt cache, fuzzy edit, guardrails, checkpoints, @-refs, subdir hints) by @everettjf in https://github.com/microclaw/microclaw/pull/342
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.53...v0.1.55

What's Changed
- feat: port hermes-agent session-search, osv-check, clarify, and SSRF guard by @everettjf in https://github.com/microclaw/microclaw/pull/335
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.52...v0.1.53

What's Changed
- feat: mid-turn message injection for interactive agent turns by @everettjf in https://github.com/microclaw/microclaw/pull/330
- fix(config): preserve YAML comments on config updates by @flyflypeng in https://github.com/microclaw/microclaw/pull/332
- fix(nix): derive web npm deps from lockfile via importNpmLock by @everettjf in https://github.com/microclaw/microclaw/pull/333
- build(deps): bump the web-non-major group in /web with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/326
- build(deps-dev): bump the web-non-major group across 1 directory with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/334
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.51...v0.1.52

What's Changed
- feat: memory & skill optimizations inspired by hermes-agent by @everettjf in https://github.com/microclaw/microclaw/pull/329
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.50...v0.1.51

What's Changed
- feat(packaging) Add initial support for snap package for Ubuntu and other snap-optional distros by @markatcanonical in https://github.com/microclaw/microclaw/pull/325
New Contributors
- @markatcanonical made their first contribution in https://github.com/microclaw/microclaw/pull/325
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.39...v0.1.50

What's Changed
- feat(weixin): enable markdown in outbound messages by @everettjf in https://github.com/microclaw/microclaw/pull/324
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.38...v0.1.39

What's Changed
- feat: concurrent mode — per-chat turn serialization + parallel tool execution by @everettjf in https://github.com/microclaw/microclaw/pull/320
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.37...v0.1.38

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.36...v0.1.37

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.35...v0.1.36

What's Changed
- ci: add automated skill review for SKILL.md pull requests by @popey in https://github.com/microclaw/microclaw/pull/311
- [codex] fix rustls websocket provider panic by @everettjf in https://github.com/microclaw/microclaw/pull/316
- stop tracking web/dist build artifacts by @flyflypeng in https://github.com/microclaw/microclaw/pull/314
- Feat support chat abort method by @flyflypeng in https://github.com/microclaw/microclaw/pull/318
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.31...v0.1.33

What's Changed
- fix(reflector): strip thinking tags from message content before LLM p… by @mszzukowski in https://github.com/microclaw/microclaw/pull/303
- Normalize malformed OpenAI tool arguments by @perrylau in https://github.com/microclaw/microclaw/pull/304
- build(deps): bump the web-non-major group in /web with 2 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/305
- build(deps): bump the cargo-non-major group with 3 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/306
- [codex] reduce release artifact size with release profile tuning by @everettjf in https://github.com/microclaw/microclaw/pull/310
New Contributors
- @perrylau made their first contribution in https://github.com/microclaw/microclaw/pull/304
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.30...v0.1.31

What's Changed
- [codex] Fix runtime config loading and invalid model fallback by @everettjf in https://github.com/microclaw/microclaw/pull/301
- [codex] fix weixin pdf downloads by @everettjf in https://github.com/microclaw/microclaw/pull/302
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.29...v0.1.30

What's Changed
- fix(web-ws): use key for websocket session json field by @flyflypeng in https://github.com/microclaw/microclaw/pull/298
- [codex] fix Feishu attachments, MiniMax tool calls, and scheduler retries by @everettjf in https://github.com/microclaw/microclaw/pull/299
- [codex] restore websocket session compatibility by @everettjf in https://github.com/microclaw/microclaw/pull/300
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.28...v0.1.29

What's Changed
- build(deps): bump sqlite-vec from 0.1.7-alpha.10 to 0.1.8-alpha.1 in the cargo-non-major group by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/288
- build(deps): bump the web-non-major group in /web with 4 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/287
- [codex] add native openclaw-weixin support by @everettjf in https://github.com/microclaw/microclaw/pull/289
- feat(web): refactor ws methods and implement dynamic sessions.list by @flyflypeng in https://github.com/microclaw/microclaw/pull/296
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.27...v0.1.28

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.25...v0.1.27

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.23...v0.1.25

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.22...v0.1.23

What's Changed
- [codex] fix web thinking tag parsing by @everettjf in https://github.com/microclaw/microclaw/pull/284
- [codex] add ACP-backed external subagent runtime by @everettjf in https://github.com/microclaw/microclaw/pull/283
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.21...v0.1.22

What's Changed
- [codex] discourage invented file paths in tool prompts by @everettjf in https://github.com/microclaw/microclaw/pull/280
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.20...v0.1.21

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.19...v0.1.20

What's Changed
- feat: improve skill scores for microclaw by @popey in https://github.com/microclaw/microclaw/pull/279
New Contributors
- @popey made their first contribution in https://github.com/microclaw/microclaw/pull/279
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.18...v0.1.19

What's Changed
- [codex] Improve Mission Control gateway/session compatibility by @everettjf in https://github.com/microclaw/microclaw/pull/278
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.17...v0.1.18

What's Changed
- [codex] keep bash tmp usage inside working dir by @everettjf in https://github.com/microclaw/microclaw/pull/271
- [codex] improve mission control bridge and web auth ui by @everettjf in https://github.com/microclaw/microclaw/pull/273
- Add Windows installer and gateway service support by @everettjf in https://github.com/microclaw/microclaw/pull/269
- [codex] add official docker image release flow by @everettjf in https://github.com/microclaw/microclaw/pull/277
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.16...v0.1.17

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.15...v0.1.16

What's Changed
- build(deps): bump the cargo-non-major group with 6 updates by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/268
- build(deps): bump @assistant-ui/react from 0.12.9 to 0.12.17 in /web by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/256
- build(deps): bump @assistant-ui/react-markdown from 0.12.3 to 0.12.5 in /web by @dependabot[bot] in https://github.com/microclaw/microclaw/pull/266
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.13...v0.1.15

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.11...v0.1.12

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.10...v0.1.11

What's Changed
- fix: parse structured Gemini text parts in OpenAI-compatible responses by @everettjf in https://github.com/microclaw/microclaw/pull/238
- [codex] add mission control websocket bridge by @everettjf in https://github.com/microclaw/microclaw/pull/236
- feat: add reusable provider presets for bot/channel overrides by @everettjf in https://github.com/microclaw/microclaw/pull/239
- Add o tel observation feat by @flyflypeng in https://github.com/microclaw/microclaw/pull/241
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.9...v0.1.10

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.8...v0.1.9

What's Changed
- Add lightweight A2A HTTP support by @everettjf in https://github.com/microclaw/microclaw/pull/228
- fix(llm): stabilize Gemini 3/2.5 tool calls and thought signature handling - fixes #210 by @Erudition in https://github.com/microclaw/microclaw/pull/230
New Contributors
- @Erudition made their first contribution in https://github.com/microclaw/microclaw/pull/230
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.3...v0.1.5

What's Changed
- feat: session-native subagents runtime v1 (spawn/list/info/kill) by @everettjf in https://github.com/microclaw/microclaw/pull/217
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.2...v0.1.3

What's Changed
- [codex] add telegram topic routing config and thread-aware chat routing by @everettjf in https://github.com/microclaw/microclaw/pull/226
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.1.1...v0.1.2

What's Changed
- Fix session id on non-default session by @RCasatta in https://github.com/microclaw/microclaw/pull/215
- nix: harden flake build and add nixpkgs upstream guide by @everettjf in https://github.com/microclaw/microclaw/pull/223
- [codex] add configurable user-agent for openai-compatible provider calls by @everettjf in https://github.com/microclaw/microclaw/pull/222
- release: add optional nixpkgs automation flow by @everettjf in https://github.com/microclaw/microclaw/pull/225
- [codex] add http chat trigger aliases and docs by @everettjf in https://github.com/microclaw/microclaw/pull/221
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.163...v0.1.1

What's Changed
- refactor: replace qwen-code preset with qwen-portal oauth provider by @everettjf in https://github.com/microclaw/microclaw/pull/214
- fix: enable Google thinking/thought support in OpenAI-compatible stream by @everettjf in https://github.com/microclaw/microclaw/pull/211
- fix: skip malformed tool calls with empty tool names by @everettjf in https://github.com/microclaw/microclaw/pull/212
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.161...v0.0.163

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.160...v0.0.161

What's Changed
- Fix scheduler timezone drift for recurring tasks by @everettjf in https://github.com/microclaw/microclaw/pull/199
- feat(setup): support qwen-code provider preset by @everettjf in https://github.com/microclaw/microclaw/pull/204
- feat(feishu): 支持表情 reaction 回复（点赞等） by @everettjf in https://github.com/microclaw/microclaw/pull/200
- fix(telegram): skip message edits during streaming in separate_message mode by @mszzukowski in https://github.com/microclaw/microclaw/pull/202
- fix(gateway): improve macOS install/status behavior by @everettjf in https://github.com/microclaw/microclaw/pull/206
- fix(windows): avoid MCP startup black console flash by @everettjf in https://github.com/microclaw/microclaw/pull/207
New Contributors
- @mszzukowski made their first contribution in https://github.com/microclaw/microclaw/pull/202
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.159...v0.0.160

What's Changed
- add flake.nix by @RCasatta in https://github.com/microclaw/microclaw/pull/191
- Journald by @RCasatta in https://github.com/microclaw/microclaw/pull/192
- mcp: strip internal microclaw keys from forwarded args by @tdonovic in https://github.com/microclaw/microclaw/pull/195
New Contributors
- @RCasatta made their first contribution in https://github.com/microclaw/microclaw/pull/191
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.158...v0.0.159

What's Changed
- fix(mcp): fix streamable HTTP transport protocol compliance by @tdonovic in https://github.com/microclaw/microclaw/pull/190
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.153...v0.0.158

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.152...v0.0.153

Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.150...v0.0.151

What's Changed
- feat(web): add skills settings management with enable/disable APIs by @flyflypeng in https://github.com/microclaw/microclaw/pull/184
- feat(cli): add microclaw upgrade self-update command by @everettjf in https://github.com/microclaw/microclaw/pull/182
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.139...v0.0.150

What's Changed
- feat(slack): support inbound image attachments by @tdonovic in https://github.com/microclaw/microclaw/pull/180
New Contributors
- @tdonovic made their first contribution in https://github.com/microclaw/microclaw/pull/180
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.137...v0.0.139

What's Changed
- feat: streaming message editing for Telegram and Matrix by @mission-deny-the-mission in https://github.com/microclaw/microclaw/pull/128
- Add multi-provider model commands and live model discovery by @mission-deny-the-mission in https://github.com/microclaw/microclaw/pull/101
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.136...v0.0.137

What's Changed
- fix(skills): fix built-in SKILL.md lookup for include_dir subdirs by @flyflypeng in https://github.com/microclaw/microclaw/pull/178
New Contributors
- @flyflypeng made their first contribution in https://github.com/microclaw/microclaw/pull/178
Full Changelog: https://github.com/microclaw/microclaw/compare/v0.0.135...v0.0.136

MicroClaw v0.0.135

MicroClaw v0.0.133

MicroClaw v0.0.130

MicroClaw v0.0.129

MicroClaw v0.0.128