Add ADR-0011: the community edition defaults to ~/.waggle-community, a
temporary name that follows the product-name decision (Q8), and copies
nothing from the commercial ~/.waggle. It supersedes point 4 of
ADR-0009 and closes the pending "separate data directory" decision.
Update the install guide, troubleshooting, capabilities, memory,
architecture, the Windows build notes, THREAT_MODEL, README,
.env.example and the changelog. Log UF-068 (Open): marketplace skills
and plugins ignore a custom data directory.
Settings > Advanced and the Vault notice showed ~/.waggle. Take the path
from DEFAULT_DATA_DIR_LABEL in @waggle/shared (ADR-0011) and update the
comments that describe the first-launch flag and skill locations.
The Tauri shell's bundled npm environment and its first-launch flag, the
server installer's --data-dir default and the NSIS installer config
fell back to ~/.waggle. Use ~/.waggle-community (ADR-0011).
The Rust shell keeps the name in DEFAULT_DATA_DIR_NAME in lib.rs and the
launcher, which uses only Node built-ins, in DATA_DIR_NAME.
packages/shared/tests/data-dir.test.ts fails when either copy, or the
marketplace's, differs from @waggle/shared.
The community edition defaulted to ~/.waggle, the data directory of the
commercial Waggle, so both editions would open the same config, vault,
waggle.db, marketplace catalog and skills. Default to
~/.waggle-community instead (ADR-0011). WAGGLE_DATA_DIR still overrides
it wherever it did; nothing is migrated from ~/.waggle.
The name is defined once as DEFAULT_DATA_DIR_NAME and
DEFAULT_DATA_DIR_LABEL in @waggle/shared, with defaultDataDir() in
@waggle/core for the sidecar, the CLI, the agent and the model caches.
The marketplace installs on its own without workspace packages, so it
keeps a copy in its data-dir.ts; the CLI help, which loads no
dependency, spells the path out and its runtime test checks it.
The marketplace security scan also flags skills that reference the new
skills folder. Tests that asserted the old default now use the
constant, and comments name the new directory.
The page showed LoCoMo scores of the old hive-mind memory engine, which
the community edition does not run, so its numbers misled. Remove the
page and its test, the /benchmarks route, the command palette entry, the
route announcement label and the color-guard baseline entry. The e2e
route check now expects /benchmarks to show the not-found page.
exceljs 4.4.0 pulled unzipper 0.10, which depends on binary and through
it on buffers 0.1.1, a package with no license grant. An npm override
moves exceljs to unzipper 0.12.3, the version whatsapp-web.js already
uses, so binary, buffers, chainsaw and traverse leave the installers.
Waggle only writes workbooks and loads them through JSZip; both still
work. The license gate drops the two exceptions it no longer needs.
ADR-0005's open item is closed: the installed Chrome or Edge first, a
downloaded Chrome for Testing only when none is found. The v1 platform
list is confirmed and leaves the pending table. ADR-0005 no longer
mentions install.sh or an npm install on the user's machine.
Records the decision not to generate marketplace.db at build time or ship
it as a release artifact (P1.6), and adds the catalog's third-party
descriptions to the pending licensing review.
A writer of the stopped app can still be finishing in the data directory
when the smoke test removes its temp root, so rmdir failed with
ENOTEMPTY after the test had passed. The temp root is disposable: a
failed cleanup now logs a warning.
The CLI sent every request to a LiteLLM proxy on :4000, which the
community edition no longer starts (it needs Python), so the CLI could
not answer with any model out of the box. ollama/<tag> models now go to
Ollama's OpenAI-compatible API directly; other models still use the
configured proxy.
The SDK reads pack manifests and starter skills next to its own module,
which in the bundle is service.js, and no build step copied them there:
installed builds showed no packs and no starter skills. build-sidecar now
copies both folders next to service.js and Tauri bundles them (UF-066).
GET /api/agent/status returned the configured model even when it had no
key, so the status bar named Claude while chats ran on the local model.
It now resolves the model the way the chat route does (UF-067).
- CHANGELOG "Removed" and the README's list of what is not in this
edition name Waggle Room tasks for coding agents and say what the Tool
Launcher still does.
- AGENTS.md lists the removal under "Already removed".
- The Room canvas no longer invites users to "launch a captured external
task"; it shows Waggle agents, groups and launched tools.
The lane drove real Claude Code, Codex, Hermes and OpenClaw installs
through Waggle Room runs, the coding-agent hooks and the hive-mind CLI,
none of which is in the community edition.
- scripts/test-windows-external-agents.ps1 and
scripts/test-windows-official-auth-canaries.ps1 (the runners and their
receipt sealing), scripts/verify-codex-tool-denial.mjs
- tests/e2e/launcher-real-hook-lifecycle.spec.ts and
tests/e2e/launcher-real-tool-lifecycle.spec.ts (Windows-only, skipped
unless the runner set them up)
- tests/integration/external-agent-collaboration.live.test.ts
- the tauri-config tests that pinned those scripts
A Launcher launch still opens the tool in the active workspace's folder,
tracks its process, streams observed output and stops it. What it no
longer does:
- set WAGGLE_SIGNAL_EMIT, WAGGLE_SIDECAR_URL, HIVE_MIND_DATA_DIR or the
Room identity variables (WAGGLE_RUN_ID, WAGGLE_ROOM_ID,
WAGGLE_DANCE_TEAM_ID, WAGGLE_SENDER_ID, WAGGLE_RUN_TOKEN). Only the
removed coding-agent hooks and the WaggleDance CLI read them. The
external-process allowlist now admits WAGGLE_WORKSPACE_ID alone;
- issue a run credential. Run tokens existed for Room workers (the dance
CLI and OpenClaw's model calls), so the registry's credential store,
the middleware's run-token bearer on /v1/chat/completions and the
proxy's per-run model scope go. While an agent run with a model is
active, the proxy still serves only the session bearer.
The run record that a launch creates in the agent-run registry stays: it
backs cancel from the Room canvas, which this change does not remove.
The `task` spec (argv templates, access modes, output dialects) and the
capability flags headlessTask, structuredProgress, resumable and
liveWaggleDance described how a Waggle Room run drove a coding agent
without its UI. Nothing runs tools headless any more, so built-in
manifests, the third-party adapter loader and detection results no
longer carry them. An adapter file that still declares `task` is
rejected like any other unknown key.
The executor router could propose Claude Code, Codex or Hermes and
dispatch the task as a Waggle Room run through POST /api/tools/run, which
is gone with the collaboration runtime. Confirming such a proposal could
only fail.
- ExecutorRegistry offers the five v1 personas; it no longer detects
external tools.
- Route proposals dispatch persona chat turns only. The memory brief and
its egress disclosure existed only for external executors, so
executor-brief.ts, the egress fields and removeFrameIds go too.
- RouteProposalCard drops the memory disclosure, "Run without memory" and
the link to an external run in the Room.
Waggle Room is not part of the community edition (decision Q2, ADR-0002).
Its headless runs of Claude Code, Codex, Hermes and OpenClaw needed the
packaged collaboration CLI from the removed @waggle/hive-mind-cli, so
POST /api/tools/run answered 503 collaboration_runtime_missing.
- Remove POST /api/tools/run (routes/external-tool-runs.ts) with its
Room fan-out, WaggleDance publishing, OpenClaw provisioning and the
memory write-back of run results.
- Remove runExternalTool (agent/external-tool-runner.ts) and
resolveWaggleRuntime; the taskkill path helper moves into the process
tracker, its only remaining user.
- Launcher: drop the hidden "Run task" composer, the Room hand-off and
adapter.runExternalToolTask. Detect, launch, watch and stop stay.
- Drop the `room` flag from the memory feature map: it only gated the
Launcher task button.
- The viewer policy no longer reads participants[].workspaceIds, a body
shape only /api/tools/run accepted.
Since phones get a bottom bar instead of the sidebar (UF-057), the shell
wait matched the hidden sidebar first and timed out at 390 px, and the
mobile journey clicked the sidebar's workspace button. The wait now takes
the first visible shell element, and the journey opens New workspace from
All workspaces, as a phone user does.
The release publishes the server archive and the desktop installers under
one tag, and the installer looks for the release named after its own
version, so both now carry 1.0.0.
The architecture guide, AGENTS.md and the threat model said releases were
built by hand or Windows-first; ADR-0009 still listed the release
workflow as future work. The Windows build guide named the missing
Settings > API Keys tab.
.env.example described Stripe, Clerk, MinIO and the team server, none of
which the community edition runs, and sent users to a Settings tab that
does not exist. The web UI no longer names a person as author, and the
installer package gets the README its files list promised.
- Key messages named a Settings > API Keys tab that does not exist; they
now point at Settings > Models (UF-061).
- The connector suggestion looked for vault.db, which never exists; it
now checks the vault for a connector entry (UF-062).
- /config and /preferences were registered with a leading slash and
never matched (UF-063).
- Pins and documents were written under ~/.waggle whatever the data
directory; they now use the server's data directory (UF-064).
- The Windows uninstaller named a Data & Privacy page that does not
exist (UF-065).
The guide now says the release workflow builds Windows on windows-latest
with NSIS and MSI, and shows how to pack the server archive and run the
installer and desktop smoke tests locally.
release.yml now has one build job with a matrix over the four release
targets (windows-x64, macos-arm64, macos-x64, linux-x64). Each leg stages
the sidecar, checks it, builds the web UI, packs the server archive for
the installer (ADR-0009), runs the installer smoke test, builds the
desktop installers (NSIS and MSI, app and DMG, deb and AppImage), runs
the packaged app until its sidecar answers /health, and uploads the
installers and archives. A pushed v* tag also gets a draft GitHub
release with a SHA256SUMS file, created with GITHUB_TOKEN only. Pull
requests on the desktop paths and manual runs build and test without
publishing, so tauri-build-pr.yml is removed.
Nothing is signed or notarized. The macOS app gets an ad-hoc signature
("-"), which needs no certificate; without it macOS reports a downloaded
Apple Silicon app as damaged.
Removed with the old pipeline, as nothing else used them: the Windows
signing handoff and signer (sign-windows-artifact.ps1 and its tests,
new-windows-signing-handoff.ps1), the pilot signing scripts
(sign-windows-pilot.ps1, apply-signing-config.mjs, signing-config.ts and
its test, sign-macos-adhoc.sh, tauri.build-override.conf.json and the
tauri:sign:* and tauri:build:win:pilot-signed scripts), the installer
certifier with its bootstrap-token helper, and the Windows release
publisher. check-sidecar-resources.mjs stays: tauri.conf.json runs it
before every desktop build and the workflow runs it before packing the
archive.
app/package.json gains tauri:build:linux. The tests that pinned the old
workflows, the certifier, the signer and the publisher are removed;
new tests pin the targets, the step order, the absence of signing and of
secrets other than GITHUB_TOKEN, and the draft-only, tag-only publish.
scripts/release-smoke.mjs runs on every release target:
- installer: installs the server archive with packages/launcher into a
temporary directory, requires a clean doctor report (only the PATH
warning is allowed), starts run on a free port with a temporary data
directory, waits for /health, stops it and uninstalls;
- desktop: starts the packaged app (an executable or a macOS .app) with
a temporary data directory, waits until its sidecar answers /health on
127.0.0.1:3333, then stops the app and the sidecar and checks that the
port is released.
The process tree is stopped through the process group on macOS and
Linux and with taskkill /T on Windows. The tests run both modes against
stand-in payloads.
The Windows installer certifier opened a WebView2 CDP port through
WAGGLE_CERTIFIER_WEBVIEW_DEBUG_PORT. The certifier leaves with the
commercial release pipeline, so the main window is built straight from
its configuration and the shipped app never opens a debugging port.
commands.md lists all 23 registered slash commands with their working
aliases and drops the FTS5 and two-mind wording. api.md explains the
bearer session token and that the API is internal, fixes the session
route parameters, lists the memory routes served through the memory
contract, the current chat stream events and the permissions body, and
removes the team, import and KVARK enterprise-pack routes.
Cover the unsigned installers, the server installer's error messages,
the desktop startup log, ports, models, memory on mem0 (including the
embedder mismatch) and the data directory. Drop team mode, the .mind
files, FTS5, docker compose for LiteLLM, the Node-on-PATH requirement
of the old desktop app and the invented cost per conversation.
Remove KVARK enterprise connectors, the 120+ package count and the
curl examples, which no longer work without the session token. Describe
Skills Hub, the marketplace scan as the code applies it, MCP Hub, the
30 native connectors, the approval modes, and the messaging channels
with the whatsapp-web.js warnings and browser download.
docs/GETTING-STARTED.md and docs/guides/getting-started.md covered the
same ground; keep one guide in docs/guides that follows the current
setup flow, the two model paths, memory on mem0 and the real keyboard
shortcuts. Drop the product tiers, the signed-Windows-only install, the
.mind files, install.sh and the external coding-agent sign-in.
The workspaces guide loses team workspaces, the .mind files, FTS5 search
and the knowledge graph, and describes virtual and local storage, the
shipped personas and what deleting a workspace removes.
List what changed for users compared with the commercial product:
removed (billing, KVARK, compliance, self-learning, team mode, old
memory engine features), changed (mem0 memory, waggle.db, opt-in
analytics, whatsapp-web.js, the server installer, the pinned Node
runtime, the built-in proxy), added and fixed, based on the ADRs and
the branch history.
Drop the personal email address. Security reports go through GitHub
private vulnerability reporting, with hello@egzakta.com as the fallback;
the code of conduct points security reports there too. Add a
supported-versions table for the first community release and update
the scope to the desktop app, the server installer and the shipped
packages.
Say plainly what Waggle is, show both install paths with a link to the
install guide, the current run-from-source steps (no build:packages),
the checks and a documentation index. The license section points to
THIRD_PARTY_NOTICES.md; the Apache-2.0 hive-mind sentence is gone with
those packages.
Describe the web UI, the sidecar, the Tauri shell, waggle.db, the
memory contract and mem0, models, the server installer and archive,
and the staging chain that builds the desktop packages and the server
archive. Drop the removed packages, KVARK, team mode, the .mind files
and the knowledge graph.
Point contributors to AGENTS.md for the rules, list the current setup
and check commands (no build:packages step, no Docker), and drop the
team-mode test setup and the stale test counts. docs/CONTRIBUTING.md
keeps a one-line pointer because the pull request template links it.
docs/install.md covers both ways to install (the Tauri desktop
packages and the server installer), the unsigned-installer warnings
on each system, first launch, the data directory, update and
uninstall. docs/memory.md describes the memory contract, its mem0
implementation, where mem0 keeps its files, recall and background
fact extraction.
Developer ID and notarization on macOS (including the bundled node and
native libraries), the owner's existing Azure Artifact Signing on Windows
with SignPath Foundation or a CA certificate as alternatives, and signed
checksums on Linux. Lists the decisions the owner has to make.
Vite inlined the two smallest Hanken Grotesk and JetBrains Mono subsets as
data: URLs. The sidecar and Tauri CSPs allow fonts from 'self' only, so
the browser blocked them and those character ranges fell back to a system
font. Fonts are now never inlined.
The launcher is now a self-contained installer with install, run, update,
doctor and uninstall. Install downloads waggle-server-<version>-<target>
.tar.gz, checks its SHA-256, unpacks it into a per-user program directory
and writes a waggle-server command that runs the tool on the archive's own
Node. Nothing is built on the user's machine and the data directory stays
WAGGLE_DATA_DIR or ~/.waggle.
scripts/build-server-archive.mjs packs the staged desktop sidecar payload
with the built web UI and a manifest. The download location waits for the
repository decision; until then install takes --from or
WAGGLE_RELEASE_BASE_URL (ADR-0009).
Removes install.sh, scripts/waggle-server.sh and their CI job, which the
installer replaces, and the bin name clash with @waggle/cli.
Sample data in tests, prompts and UI placeholders uses Alex Morgan
instead of the maintainer's name; decision attributions say "owner".
Personal details in two test fixtures are replaced. Local paths
(D:/Projects/...) and paths into the maintainer's planning repository
are reworded. Clone and install URLs in CONTRIBUTING.md and the getting
started guide use placeholders until the public repository location is
decided (Q9). The Benchmarks page no longer names the maintainer's
hive-mind repository, and the upstream fixes log names the main Waggle
repository instead of an internal git host.