Skip to content

Changelog

Unreleased

  • No-code HTTP connections accept http:// base URLs and warn that traffic is not encrypted.
  • Add development-only private origins to the Docker development platform: weave platform up --allow-private-origin ORIGIN (repeatable) lets HTTP connector actions and signed webhooks reach a local http:// test service at that exact origin on the installation's own egress network. The API reads the approved entries from the read-only file named by WEAVE_PRIVATE_ORIGINS_FILE and refuses to start when that file is malformed, a symbolic link, or writable by other users.
  • Parse WEAVE_HTTP_PRIVATE_NETWORKS, WEAVE_MAIL_PRIVATE_NETWORKS, and the PostgreSQL private and plaintext network settings strictly at startup: a CIDR with host bits set, or more than 128 networks, in any of them stops the API. Before, the HTTP and mail settings accepted a CIDR with host bits set and failed the requests that used it, and the HTTP and PostgreSQL settings had no limit. Reach is otherwise unchanged.
  • Connection test answers carry a new encrypted field. CLIs and SDKs older than this release cannot read weave connections test answers for HTTP connections from an upgraded server; upgrade the CLI and SDK together with the server.

0.1.0a14

  • Add a detached Docker development platform through weave platform up, reusing owned databases, identity setup, workspace creation, and user roles. Keep the foreground API mode available for existing installations.
  • Prepare an explicit Developer ID signing and notarization path for macOS releases. Apple credentials are required; ordinary builds remain ad-hoc signed.
  • Pin the Agentic and Files 0.1.6 packages to core 0.1.0a14 without changing their execution behavior or the database schema.

0.1.0a13

  • Retry explicit platform capacity rejections while a worker reads its task context or credentials, using the current validated lease. Reject late responses after cancellation, renewal failure, or expiry; unrelated HTTP errors and ambiguous network failures are not replayed.
  • Classify an Agentic preparation timeout as not started when its own time budget expires before provider execution. Failures after provider execution begins retain their ambiguous outcome classification.
  • Publish Agentic and Files worker packages 0.1.5 pinned to core 0.1.0a13. Files has a dependency update only. The worker protocol and database schema remain unchanged from alpha12 (0030_worker_presence).

0.1.0a12

  • Add environment-scoped deployment targets, desired state, observations, immutable approval plans, fenced runner jobs, and explicit reconciliation for uncertain external effects. Provider credentials remain on an outbound runner with local destination and image allowlists.
  • Add Operations in Studio and the CLI for reviewed container deployment changes. Support trusted Compose deployment and bounded existing Kubernetes and Azure Container Apps changes; provisioning cloud clusters is separate.
  • Report dated worker contact and capacity, with revision-checked drain and resume controls that preserve active leases.
  • Let Lumi explain explicitly selected Operations records using bounded, permission-checked context. Deployment changes retain their separate plan, approval, and application process.
  • Reject unsupported component and destination changes before plan approval. Mark stopped or stalled compatibility monitors unhealthy so container supervisors can replace the affected API process.
  • Guide target registration with named application accounts, downloadable nonsecret runner setup, observed-resource import, explicit resource capacity, and admitted worker release selection.
  • Allow reviewed Compose plans to restart stopped workers, checking readiness after the restart instead of during the dry run.
  • Keep worker lease renewal retrying explicit capacity rejections while the current lease remains valid, including while a completed task awaits its acknowledgment. Ambiguous transport failures never replay the task handler.
  • Keep invalid inspector edits and unapplied AI profile changes visible before saving, publishing, or switching views. Edit workflow AI profiles in a dialog, with multiline prompts and direct connection-slot selection.
  • Organize editor commands into aligned rows and step settings into purposeful sections. Keep palette search visible and explain data sources through keyboard-accessible information controls. Filter source modes and formulas by the receiving field's type while preserving imported expressions for repair.
  • Separate reusable project actions from creating a new API action, with a direct connection entry point and contextual guidance for local authoring.
  • Keep Studio pages and editor assets loading while platform requests are busy, and retry temporary session-read capacity rejections within a bounded budget.
  • Use standard OpenAI and Anthropic endpoints automatically, retain explicit Azure configuration, and group advanced model settings separately. Check model-setting combinations with the canonical local validator before applying them; configuration validation does not claim live provider readiness.
  • Add migrations 0029_deployments and 0030_worker_presence, with explicit scoped Operations roles. Existing workflow roles gain no deployment authority.
  • Publish worker packages 0.1.4 pinned to core 0.1.0a12 and desktop alpha12. Keep the verified alpha10 Azure acceptance history separate from this release.

0.1.0a11

  • Guide AI provider connections, Lumi settings, and workflow AI profiles through focused steps with review before creation or application. Keep unfinished profile changes local and apply the profile and connection slot atomically.
  • Let authors explicitly pass available earlier AI results as structured context within one workflow execution, using existing compiler-checked expressions.
  • Document the verified alpha10 Azure application-token, AI workflow, Lumi, and HTTP acceptance separately from unverified interactive Entra sign-in.
  • Correct browser selection tests to wait for the insertion reveal before checking pointer reachability.
  • Publish worker packages 0.1.3 pinned to core 0.1.0a11. No server behavior or database migration changes; the schema remains 0028_files.

0.1.0a10

  • Expose Lumi configuration through the paired Studio host and clarify the separate administrator, worker-admission, and deployment responsibilities.
  • Let explicit canvas selection and structural edits supersede pending initial fitting. Keep automatically added empty fields from stealing keyboard focus.
  • Use the Lumi mascot in Studio and the illustrated documentation.
  • Make compiler scope-parity tests reject inconclusive resource-limit results and isolate their timing allowance from production validation budgets.
  • Publish worker packages 0.1.2 pinned to core 0.1.0a10. No database migration is added; the server still requires schema 0028_files.

0.1.0a9

  • Add guided Studio setup for AI provider connections, named connection selection for Lumi, and independent workflow model profiles. Keep model credentials in operator-managed secret storage and show scoped setup permissions explicitly.
  • Keep incomplete LLM profile edits open until required properties are present.
  • Add renewable OAuth2 client credentials for independently deployed Agentic workers, with bounded token acquisition, expiration-aware caching, scoped API origins, cancellation cleanup, and mounted-secret rotation.
  • Explain AI workflow execution, Lumi proposal review, and configuration roles with step-by-step guides and accessible SVG diagrams.
  • Publish worker packages 0.1.1 pinned to core 0.1.0a9. No database migration is added; the server still requires schema 0028_files.

0.1.0a8

  • Add verified file references, resumable bounded transfers, scoped file roles, CLI upload/download commands, Python SDK streaming helpers, worker lease checks, and human-task attachments with current claimant/revision checks. PostgreSQL stores file chunks separately from run history; retained runs pin their files until the run is purged.
  • Add independently deployed FTP, FTPS, SFTP, SharePoint/OneDrive and Google Drive workers for list, metadata, download, write, move and delete operations. Cloud account acceptance remains environment-specific; change-feed triggers are not included. FTP/FTPS/SFTP require a server-isolated account root. FTP/FTPS write and move are disabled by default because the protocol cannot guarantee atomic refusal to replace a concurrently created destination; enabling them requires an explicit connection policy and a separate worker policy.
  • Add contains/not-contains and membership comparisons, versioned decision tables, and schema-guided Studio editors for rules and AI steps.
  • Add workflow llmProfiles and Agentic workers with explicit provider, model, reasoning pattern, endpoint policy and budgets. Pin the profile into compiled execution, enforce output guards, and reject classified results before values or their hashes enter durable state, simulation or reconciliation.
  • Add Lumi as a separate environment-configured assistant, using an authenticated private gateway. Studio shares only selected context, validates proposed definitions, and requires explicit Apply with stale-draft protection and Undo.
  • Improve Studio's step configuration, input mapping and human-task authoring; render decisions and parallel branches as separate lanes; apply valid field edits immediately; keep invalid drafts local; add semantic drag/drop and reachable node actions.
  • Add schema revisions 0026_decision_tables, 0027_lumi_configuration, and 0028_files. Upgrade through the documented maintenance procedure before starting the new server. Older grants do not automatically grant file or Lumi access.
  • Package independent workers separately so Agentic's Python 3.13 requirement does not change the server's Python 3.12 support.
  • Keep worker completion and failure retries alive after explicit capacity rejections while the lease and task deadline remain valid. Direct transport calls retain bounded retries; ambiguous outcomes are never retried automatically.

0.1.0a7

  • Add saved platforms shared by the CLI, Studio, and desktop app: a non-secret profiles.json in the system configuration folder (WEAVE_CONFIG_HOME overrides it), written atomically with private permissions. Tokens stay in the system credential store, or in an explicit private file.
  • Add weave auth setup SERVER. It checks the server, shows its sign-in settings for review, signs you in through the system browser (PKCE with a loopback callback) or a device code, lets you choose a workspace, and says what was saved and which command to run next. Explicit flags, --yes, and --output json keep it usable in scripts without prompts.
  • Add weave auth profiles, use, remove, workspace, and login --switch-account. weave auth status exits 1 when you are not signed in.
  • Add the public GET /api/v1/client-configuration endpoint, configured with WEAVE_CLIENT_SIGN_IN and WEAVE_DISPLAY_NAME. Clients treat it as a proposal: you review it once, the issuer, client, and endpoint origins are pinned, and a later change is refused with WV-PROFILE-CHANGED.
  • Check server addresses before connecting: HTTPS except on loopback, no redirects, bounded responses, and no link-local, metadata, NAT64, or 6to4 destinations.
  • Rebuild the Studio connection wizard: work locally or connect to a platform, pick a saved platform, check the connection, review the sign-in settings, sign in through the system browser or a device code, and choose a workspace. Studio reconnects on reopen and never asks for your identity-provider password. Local authoring still needs no sign-in.
  • Keep desktop pairing across restarts and save Studio exports through the system download handler.
  • Add no-code REST integrations on the built-in weave-http@2.0.0 connector: weave connector http-action, weave connector import-openapi --target builtin, guided weave connections create flags, connector descriptors, and the Studio New API action builder and OpenAPI import. Publishing now rejects an HTTP action whose configuration contradicts its method, with a pointer to /spec/implementation/config.
  • Add weave platform integrations and weave platform secret so a local platform can run built-in HTTP actions with secrets stored by handle.
  • Make the Call an action step usable end to end in Studio: choose an action and a connection, fill inputs through generated forms, map data, validate without a platform, save, reopen, and import or export YAML and JSON.
  • Redesign the Studio editor: one primary command that follows the lifecycle, readable zoom with Fit all and Tidy layout, Delete with Undo, Decision conditions as rule rows with Otherwise, problems shown on the steps, an inspector with Apply changes and Discard, and a docked simulation that highlights the live path.
  • Rework the Studio shell: Build and Operate navigation, a Home page that leads with what needs you, toasts with Undo, drafts kept on this computer with autosave, a run detail page for operators, consistent status names, and accessibility fixes for contrast, focus, and side panels at narrow widths.
  • Register port-free loopback callbacks for the local Keycloak weave-cli client, and repair existing local platforms on weave platform start, so browser sign-in works on any loopback port.
  • Keep in-process native workers running when they claim work while the platform is still opening effects; the dispatcher restarts a failed worker with backoff and reports not ready after three consecutive failures.
  • Replay in-process native worker claims, renewals, and settlements that the platform rejects for capacity, with the same policy as remote workers, so a busy database no longer exhausts a task's retries.
  • Accept JWKS signing keys that omit alg, as Microsoft Entra ID publishes them, for the allowed algorithm of their key type only (RSA for RS256, P-256 for ES256).
  • Retry Studio reads, and changes that carry an idempotency key, when the platform turns them away for capacity (WV-OPERATION-CAPACITY, WV-REQUEST-CAPACITY), honoring Retry-After for up to four attempts.
  • Show only the logo in the CLI help banner.
  • Bump the Python package and browser Studio bundle to 0.1.0a7 and the desktop product to 0.1.0-alpha.7; the native internal installer counter is 0.1.7.
  • Rewrite the documentation around four learning paths, with new guides for signing in, no-code REST integrations, and readers coming from BPM tools.

Verified on macOS against a real local Keycloak 26.7.4: CLI sign-in and workspace selection with a private credential file, a no-code HTTP action running on a local platform, the Studio sign-in journeys with the macOS Keychain, and sign-in with a code, reconnection after a restart, and export in a locally built macOS desktop app. In the desktop app, drafts kept on this computer last only until the app quits; use Save to file to keep a copy. Microsoft Entra ID and other OpenID Connect providers are supported through configuration but not verified; Entra-shaped keys are covered by unit tests only. Credential storage on Windows and Linux is not verified. Existing --auth-config connection files keep working. Schema revision remains 0025_run_lifecycle; no new database migration is added.

0.1.0a6

  • Fix the macOS desktop outer application resource seal with explicit ad-hoc signing, replacing the damaged alpha5 macOS installers.
  • Add strict codesign verification of the outer bundle, native executable, and frozen host, then repeat it against the application inside the read-only DMG before collecting release assets.
  • Keep hardened runtime with the PyInstaller library-validation entitlement and smoke-test the host after signing, before collecting installers.
  • Bump the matching Python host/browser bundle to 0.1.0a6 and desktop product to 0.1.0-alpha.6; the native internal installer counter is 0.1.6.
  • Update installation guidance to use matching alpha6 artifacts and retain the browser fallback. Do not use the alpha5 macOS downloads.

macOS bundles are ad-hoc signed, not Developer ID signed and not notarized. Resource-integrity verification does not establish Gatekeeper acceptance or publisher trust. Manual approval may still be required by macOS and must comply with local policy; no quarantine-removal workaround is recommended. Windows publisher signing is also not provided. Schema revision remains 0025_run_lifecycle, as introduced by alpha5; no new database migration is added.

0.1.0a5

  • Add the local Angular Studio with a branded workflow editor, structured canvas, source editing, API profiles, run inspection, human tasks, and email conversations.
  • Add weave studio, browser pairing, a same-origin API bridge, and a versioned, checksummed optional asset bundle that runs without Node.js.
  • Include the Studio host in the CLI installer's verified dependency closure; install the matching browser bundle separately with its published checksum.
  • Add native human tasks with scoped assignments, schema-driven decisions, durable waits, and explicit operator pause/resume controls.
  • Add generic SMTP/IMAP email integration, conversation replies, fenced submissions, inbound receipts, and authorized correlation to workflow runs.
  • Add identity/workspace discovery and separate task and email permissions.
  • Add exact business/correlation-key and status searches across independent runs, terminal-run archive/restore, and separately authorized permanent deletion with dependency checks, audit receipts, and protection against idempotent recreation.
  • Add a guided Studio connection setup with reviewed CIAM configuration, cancellable device/PKCE sign-in, connection checks, and authorized workspace selection.
  • Add scoped people/access administration through the API, SDK, CLI, and Studio, keeping CIAM account management separate from Weave identity links and grants.
  • Add typed configuration for every workflow step, recursive expression editing, integration action discovery, workflow settings, and a centered Start/End canvas.
  • Add a Tauri desktop wrapper, branded drag-to-Applications macOS disk image, and native installer build matrix. Desktop artifacts are unsigned.

These changes require the matching server, migrations through 0025, CLI, and Studio build. Back up the database and follow the upgrade guide. This remains an alpha; desktop signing/notarization and live CIAM/mail/cloud validation are separate deployment acceptance steps. Native human tasks use IR v1alpha2; workflows that do not use them retain the prior IR. The earlier intermittent queue timeout is not claimed fixed by this release.

0.1.0a4

  • Add weave platform commands for an owned local developer installation: prerequisite checks, setup, foreground startup, readiness, first run, token refresh, and stop/resume with retained data.
  • Show stage progress and terminal-only animation during setup; keep JSON output clean and offer WEAVE_NO_ANIMATION=1 for plain terminal messages.
  • Introduce Lumi, the firefly guide, in ASCII CLI help and editable documentation SVGs.
  • Export compiled workflow graphs as terminal text, Mermaid, or standalone SVG.
  • Add opt-in same-origin Swagger UI at /docs and the actual contract at /openapi.json; authenticated operations retain their existing grants.
  • Generate the complete public API reference from the same OpenAPI contract.
  • Add step-by-step SDK, YAML/Python workflow, custom inbound/outbound connector, API playground, local platform, and remote deployment tutorials.
  • Redesign the documentation diagrams with human labels, vector icons, and explicit flow explanations; annotate deployment command blocks with their purpose.

This remains an alpha. The local stack is a developer installation, not a production availability claim. Cloud and live-provider validation remain specific to the target environment. The earlier intermittent queue timeout remains unresolved; no retry or runtime scheduling policy is changed here.

0.1.0a3

  • Render the official woven-W logo as plain ASCII in CLI help, with a compact layout for narrow terminals and clean machine-readable output.
  • Add weave init DIRECTORY for a validated offline starter with a runnable simulation and instructions for editing, recompiling, and simulating again.
  • Add weave docs [TOPIC] and clearer help linking authoring, platform startup, worker deployment, and configuration to their guides.
  • Publish branded, searchable MkDocs documentation on GitHub Pages with light and dark appearances, copyable commands, and complete task-based navigation.
  • Explain the platform startup path, service responsibilities, readiness checks, first use, shutdown, and restart; distinguish optional container stages.
  • Check the documentation site in CI and retain source attribution in stylesheets.

No runtime engine, database migration, authentication, or worker retry policy changes are introduced. The intermittent queue timeout recorded in 0.1.0a2's release notes remains an unresolved observation; this release does not claim to fix it.

0.1.0a2

  • Add a user-local CLI installer with verified release assets, hash-locked client dependencies, isolated environments, safe upgrades, and managed uninstall.
  • Add ASCII branding, grouped plain-language help, nested command help, and weave --version, while preserving machine-readable command output.
  • Reorganize onboarding around local authoring, an existing API, and running the platform; explain authentication, scope, expected results, and recovery steps.
  • Add illustrated CLI, Kubernetes, AWS, Azure, and Google Cloud deployment guides.
  • Include installer assets, deployment templates, and the complete public documentation in the source distribution; verify installation during checks.

No database migrations or workflow-language changes are introduced in this release.

0.1.0a1

  • Typed workflow schemas, expressions, bounded compilation and canonical artifacts.
  • Draft/publication/activation lifecycle, durable PostgreSQL runtime, workers, waits, signals, schedules, simulation, history/replay and incident controls.
  • Native PyFly API, Python SDK/CLI, provider-neutral identity boundaries and local grants.
  • HTTP/webhooks, PostgreSQL, Kafka, integration outbox, connector authoring and provider inbox.
  • Teams personal-bot text, WhatsApp Cloud API text/template/status scope, and Telegram webhook text integration with local fixture/backend verification.
  • OpenAPI 3.1 connector generation with bounded HTTP profiles and offline diagnostics.
  • Operational capacity policies, compatibility checks, retention controls and optional telemetry.
  • Container packaging, remote worker deployment, and backup/restore guides.
  • Editable SVG diagrams and Apache 2.0 source attribution.

See current capabilities for integration scope and verification boundaries. Messaging connectors have local protocol verification; live provider account provisioning and message delivery require separate validation.