Skip to content

chore(deps): update fullsend to v0.52.0 - #305

Open
hermes-renovate[bot] wants to merge 1 commit into
mainfrom
renovate/fullsend
Open

chore(deps): update fullsend to v0.52.0#305
hermes-renovate[bot] wants to merge 1 commit into
mainfrom
renovate/fullsend

Conversation

@hermes-renovate

@hermes-renovate hermes-renovate Bot commented Oct 19, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Note

This PR body was truncated due to platform limits.

This PR contains the following updates:

Package Type Update Change Pending
fullsend-ai/agents minor 0.45.0 → 0.52.0 0.54.0 (+1)
fullsend-ai/fullsend workflow minor v0.45.0 → v0.52.0 v0.54.0 (+1)

Release Notes

fullsend-ai/agents

v0.52.0

Compare Source

Synced with fullsend v0.52.0

What's Changed

Full Changelog: fullsend-ai/agents@v0.51.0...v0.52.0

v0.51.0

Compare Source

Synced with fullsend v0.51.0

What's Changed
New Contributors

Full Changelog: fullsend-ai/agents@v0.50.0...v0.51.0

v0.50.0

Compare Source

Synced with fullsend v0.50.0

What's Changed
New Contributors

Full Changelog: fullsend-ai/agents@v0.49.0...v0.50.0

v0.49.0

Compare Source

Synced with fullsend v0.49.0

v0.48.1

Compare Source

Synced with fullsend v0.48.1

v0.48.0

Compare Source

Synced with fullsend v0.48.0

v0.47.1

Compare Source

Synced with fullsend v0.47.1

v0.47.0

Compare Source

Synced with fullsend v0.47.0

v0.46.1

Compare Source

Synced with fullsend v0.46.1

v0.46.0

Compare Source

Synced with fullsend v0.46.0

fullsend-ai/fullsend

v0.52.0

Compare Source

GitLab support leaves its trial: the first release GitLab projects can pin. Until now, GitLab installs ran Fullsend from main: the poll, webhook dispatcher and agent jobs pulled ghcr.io/fullsend-ai/fullsend-runner:dev, which is rebuilt from every main change, so an OpenShell or tool bump reached every GitLab project as soon as it merged, before any release was tested with it. From this release the GitLab CI scaffold renders the runner image tag from gitlab.fullsend_ref, the same ref that selects the CLI the jobs install, so a GitLab project runs a tested, released runtime the same way a GitHub project pinned by commit and tag does:

gitlab.fullsend_ref Runner image
A release, vX.Y.Z (the default: install writes the CLI's release here) X.Y.Z, the image published for that release
Unset, with a locally built CLI latest, which moves only when a release is published
A branch such as main, or a commit SHA dev, rebuilt from main: an opt-in for maintainers and early testers, not supported for production

Upgrade: set gitlab.fullsend_ref to v0.52.0 and re-run fullsend repos install. The same converge run re-renders the image tag in all three CI templates, and fullsend repos status reports a template whose tag does not match the pin as drift. Pins from v0.46.0 onwards, whose templates still hardcode :dev, are rewritten to their own release image. To run a release's image, pin the release tag, not its commit SHA: a SHA always resolves to dev. See Runner image version in the GitLab guide.

fullsend lint is stricter about validation_loop.preflight_check. A value shaped like a path, such as scripts/common-preflight.sh, is now an error instead of a warning (ADR 0128): the referenced file is not delivered to a consumer's checkout, so the check can only fail at run time. Write the command itself instead.

Release notes now cover the agents too. Every release page ends with an Agents changes link to the fullsend-ai/agents release for the same tag, which now lists the agents pull requests merged since the previous release.

Changelog
Breaking Changes
Bug Fixes
Agents changes

The fleet agents ship in fullsend-ai/agents v0.52.0; its release notes list the agents changes in this release.

v0.51.0

Compare Source

Signed commits for the code and fix agents on GitHub. If your repository has a ruleset that requires signed commits, the code and fix agents can now satisfy it. With GITHUB_SIGNED_COMMITS set to "true", their post-scripts no longer git push the commits built in the sandbox; they recreate each commit through the GitHub Git Data API with the App installation token, so GitHub signs them, they show as Verified, and they pass required_signatures rulesets. It is off by default ("false" keeps git push). To turn it on, give the agent a custom harness that builds on the fleet one and sets the value in its GitHub overlay, for example .fullsend/code.yaml (and the same for fix.yaml):

base: https://raw.githubusercontent.com/fullsend-ai/agents/<SHA>/harness/code.yaml#sha256=<sha256sum>
overlays:
  - when: 'runtime.forge == "github"'
    env:
      runner:
        GITHUB_SIGNED_COMMITS: "true"

GitHub makes the App's bot user the author of each published commit; the identity the agent committed with is kept in an Authored-by: trailer, which the fix agent uses to tell its own commits from code-agent and human ones. Each file, tree and commit is one API request, so a very large commit can hit GitHub's secondary rate limit. GitLab keeps git push. See Signed commits in the agents repository's docs/code.md.

fullsend lint checks your .fullsend configuration with the same validation agent runs use, including every agents[].trigger, and exits 1 on an error, so a typo in a trigger fails your pull request instead of every agent run. With --offline it now resolves built-in agents from the cache an earlier online run filled; run that warming run with the same --forge as the offline one. See Checking a trigger before you merge in the CEL triggers reference.

Breaking changes. roles in config.yaml no longer decides which agents run; turn an agent off with agents[].enabled: false. validation_loop.preflight_check must now be a credential-safe command (an allowlist), and a check that times out no longer hangs on processes it forked. On GitLab, the legacy upgrade paths are gone (pipeline-variable dispatch, the old webhook-dispatch scaffold migrations, the poll-state variable migration and unversioned rotation state), so projects must pin fullsend_ref v0.46.0 or later before they upgrade.

GitLab. Vertex inference can now use an existing Workload Identity provider that the installer does not own, such as one a platform team runs for a whole GitLab instance, with optional service-account impersonation: set the new optional inference.wif_audience and inference.service_account keys; leaving them unset keeps today's behaviour. Scaffold installs that go through a fork now open their merge requests correctly, fullsend repos status reports whether the webhook fast path is ready and diagnoses drift, comment edits no longer trigger the dispatcher, and Analyst changes-requested notes survive polling.

Changelog
Breaking Changes
Features
Bug Fixes
Refactoring

v0.50.0

Compare Source

Choose when the fleet agents run, from config.yaml. A CEL trigger on the review, retro, triage or prioritize entry in .fullsend/config.yaml now decides when that agent runs on GitHub, in place of the dispatch workflow's built-in rules. For example, to stop the automatic review on every pull request but keep /fs-review on demand:

agents:
  - name: review
    trigger: >-
      event.transition.kind == "comment_added"
      && has(event.entity.linked_change_proposal)
      && has(event.transition.comment.command)
      && event.transition.comment.command == "/fs-review"
      && event.actor.kind == "human"

Repositories without a trigger keep today's behaviour, so nothing changes unless you opt in. The trigger is read from your default branch, so a pull request cannot change its own routing. An invalid trigger makes every agent run in the repository fail with invalid agent config, so check the expression before you merge it. A harness trigger: never changes when a built-in agent runs; copy it into config.yaml to opt in. Triggers for the code and fix agents are coming in a later release. See Choosing when review runs, Choosing when retro runs and Choosing when triage and prioritize run in the customizing-agents guide.

Permission checks on CEL dispatch are now per harness: agents whose harness sets role: triage or role: review can be started by users with triage access, while every other role still needs write. Claude runs now set CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 on every route, including Vertex, a harness with its own ANTHROPIC_BASE_URL and runs with security hooks off; before, only the runner-managed gateway route set it. Nothing for you to change.

On GitLab, fullsend run adds the configured forge host to the attached credential profile, so self-managed hosts work with the stock GitLab providers. fullsend repos install now applies runner tag changes even from a build without a release ref, keeps blank lines when it rewrites .gitlab-ci.yml, and resolves extends: and !reference when it scans trigger jobs. Runner VMs refuse job images that pin OpenShell 0.0.x with a clear error instead of downgrading the host. fullsend agent new now generates agents that read the forge-neutral work item URL.

Changelog
Breaking Changes
Features
Bug Fixes

[!IMPORTANT]
✂ PR body was truncated to here.

Configuration

📅 Schedule: (UTC)

  • Branch creation
    • At any time (no schedule defined)
  • Automerge
    • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about these updates again.

  • If you want to rebase/retry this PR, check this box

This PR has been generated by Mend Renovate CLI.

@hermes-renovate
hermes-renovate Bot requested a review from a team October 19, 2026 22:25
@coderabbitai

coderabbitai Bot commented Oct 19, 2026 •
edited
Loading

Copy link
Copy Markdown

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

Run configuration

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • Trigger review

Tip

🎯 Customize your risk assessments

Guide how CodeRabbit assesses merge risk and security architecture risk. Highlight sensitive changes and explain your security boundaries using Merge Risk Instructions and Security Risk Instructions in your review settings. Security architecture risk assessments require the Advanced plan or above.

✅ Automatic approvals, with your rules

Let CodeRabbit approve pull requests once required threads are resolved, pre-merge checks pass, and your risk limits are met. Enable Automatic approvals in review settings, with overrides for teams or file paths.

@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.46.0 chore(deps): update fullsend to v0.46.1 Oct 20, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.46.1 chore(deps): update fullsend to v0.47.0 Oct 21, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.47.0 chore(deps): update fullsend to v0.47.1 Oct 21, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.47.1 chore(deps): update fullsend to v0.48.0 Oct 26, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.48.0 chore(deps): update fullsend to v0.48.1 Oct 27, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.48.1 chore(deps): update fullsend to v0.49.0 Nov 1, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.49.0 chore(deps): update fullsend to v0.50.0 Nov 2, 2026
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.50.0 chore(deps): update fullsend to v0.51.0 Nov 6, 2026
@hermes-renovate
chore(deps): update fullsend to v0.52.0
df34e6f 
Signed-off-by: hermes-renovate[bot] <286339580+hermes-renovate[bot]@users.noreply.github.com>
@hermes-renovate hermes-renovate Bot changed the title chore(deps): update fullsend to v0.51.0 chore(deps): update fullsend to v0.52.0 Nov 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

0 participants