Skip to content

Tale v0.5.72

Choose a tag to compare

View all tags
@github-actions github-actions released this 04 Oct 21:17
· 363 commits to main since this release
v0.5.72
43f29ec
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Highlights

  • Chat can use OpenAI's GPT-6 models. GPT-6 Astra, GPT-6 Sol, GPT-6 Luna and GPT-6.1 Sol are
    offered through an OpenAI API key or environment-variable credential, and Tale calls Astra and
    GPT-6.1 Sol through the Responses API, the only way they make tool calls. Project agents and
    automations run those two on Codex; another runtime now says so instead of refusing without a
    reason. Subscription credentials are marked Tasks and automations only in Settings, and the
    chat model list names the subscriptions it leaves out.
    Run the GPT-6 models
    · #4181
  • A standing task that stays in Backlog or To do while its agent works is retried
    automatically after a retryable failure, as long as its status, assignee, archive state and
    review stay unchanged. Before, the task kept reporting a pending retry and held an idle agent.
    Triggers
    · #4187
  • A workflow step whose sandbox workspace is being destroyed now fails at once with the reason,
    instead of waiting up to two hours for room and then running in a fresh, empty workspace.
    Sandboxes
    · #4144
  • A data-subject erasure completes only once the sandbox service confirms that the person's agent
    workspaces are deleted. Until then its receipt reads Partial, and Retry finishes it
    later. The audit log's Sandbox workspace deleted event also waits for that confirmation.
    Data subject requests
    · #4094
  • Sandboxes hold up better under sustained load. They release finished command output,
    disconnected readers and stalled streams, cap held-open input at 8 MiB, rotate the logs of
    nested Docker containers, replay long runs faster and no longer repeat cleanup work that used
    CPU. A package-cache volume deleted shortly after it was prepared comes back writable, so
    package installs no longer fail with permission errors.
    #4174,
    #4182,
    #4184,
    #4185,
    #4188,
    #4154
  • The sandbox runtime updates all nine bundled agent runtimes (Claude Code 2.1.285, Codex 0.160.0,
    Cursor 2026.10.01, Gemini CLI 0.62.0, Hermes 0.19.0, OpenClaw 2026.9.8, OpenCode 1.18.34,
    Pi 1.0.1 and Qwen Code 0.24.7) on Node 24.21.0. Qwen Code no longer starts the Playwright MCP
    server on every turn; the server starts with the first browser tool call.
    Harnesses
    · #4140
  • Fixes you can see in the app:
    • Add credential shows a loading state while the provider or connector catalog loads,
      instead of reporting that the deployment ships no provider files.
      #4161
    • Home drops a deleted or archived project's tasks and chats without a reload
      (#4162), and shows the Draft badge for
      a draft of literal code such as <Button />
      (#4204).
    • Comments written by automations are labelled as automation instead of "Deleted agent".
      #4164
    • A recorded approval decision no longer answers with an error when waking the run fails
      afterwards, and the approval card offers Try again when the approval cannot be loaded or
      keeps Approve and Reject when a decision was not recorded.
      Approvals in workflows
      · #3881
    • Adding teams to a project's audience one after another keeps every team
      (#4197), and a renamed team shows its new
      name in the open detail panel and editor
      (#4193).
    • Product Create stays disabled while the product is saved, so a repeated click cannot send
      a second request (#4198), and the product
      list formats prices in the reader's locale
      (#4205).
    • A queued or failed outbound mail and a chat message keep the attachment bytes they list, so
      deleting the file no longer breaks the send or a later chat turn.
      #4157
    • A website's scan-failure alert no longer comes and goes with the page filter
      (#4135), and the arrow, Home and End keys
      select the focused option of a segmented control such as that filter
      (#4163).
  • The front pages of the website, the documentation and the UI documentation are reworked, and
    the pricing table names the data processing agreement as the service only Enterprise includes.
    #4139,
    #4148
  • Operators can sample backend HTTP and worker spans through the existing Sentry integration. It
    is off by default.
    Observability

Upgrade notes

  • The backend applies database migration 0149 when it starts. It adds the state that standing
    task retries need and an index on task activity, and it retires the pending retries of failed
    standing-task runs from earlier versions without starting catch-up work; their history stays.
    Building the index can lengthen the first start on an instance with a long task history.
  • Two optional environment variables, documented in the
    environment reference,
    change nothing until you set them:
    • SANDBOX_DOCKER_WORKLOADS limits which workloads may start Docker inside the sandbox when
      the deployment enables it. The default, project,workflow, keeps today's behavior; project
      runs workflow agents without Docker. It applies to sessions created afterwards. The
      repository's compose.yml and tale deploy pass it to the sandbox service; a hand-written
      Compose file or manifest must pass it as well.
    • BACKEND_SENTRY_TRACES_SAMPLE_RATE, 0 by default, samples backend HTTP and worker spans.
      It needs SENTRY_DSN and a destination that accepts Sentry transactions. Recreate the
      backend API and worker services after changing it.
  • Without an explicit SANDBOX_AGENT_MEMORY, a sandbox session without Docker inside now
    defaults to 4g of memory and a session with Docker to 8g, even when other sessions in the
    deployment use Docker.
  • The sandbox runtime image grows to about 6.0 GB on amd64 with the refreshed agent runtimes.
    Allow for the larger pull and disk use.
  • Update the sandbox service and connected sandbox devices together with the platform. Older
    versions cannot confirm workspace deletion, so erasure receipts remain Partial until those
    services and devices are updated; retrying alone cannot fix the missing confirmation.
  • On Kubernetes, sandbox Pods that are still Pending count against admission, and maintenance
    removes a Pod left Pending past its startup deadline plus 60 seconds while keeping its
    workspace volume. A deleted workspace counts as erased once its volume is handed to the storage
    provisioner; with a Retain reclaim policy, its files stay until you remove the volume.
    Kubernetes
  • Known limitations: bytes that a queued mail or a chat message keeps are not removed when that
    mail or chat ends, and a chat attachment can keep a document's bytes after a retention
    destruction (#4208). GPT-6 models in chat
    do not carry their reasoning across tool rounds within a turn.
  • The API contract stays at 3.15.0.

API contract changes

None in this range. The contract stays at 3.15.0: 139 operations.

What's Changed

Full Changelog: v0.5.71...v0.5.72

Contributors

yannickmonney
Assets 8
Loading