[dependabot]

Bumps fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml from 0.43.0 to 0.46.1.

Release notes

Sourced from fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml's releases.

v0.46.1

Highlights

Agent token minting now retries the OIDC token fetch for about a minute instead of giving up after about 30 seconds, so a short outage of the CI OIDC endpoint no longer fails the run (#8276). The sandbox image updates the pi-inference-gateway extension to v0.1.1, which can read inference-gateway.json from its own directory and discovers both OpenAI- and Anthropic-format model lists; it stays installed but not loaded until the gateway route ships (#8282). New docs cover running pi locally through an inference gateway, including per-model settings (#8274, #8278).

---

Changelog

Bug Fixes

• 8ad73777e2d06c62c6bc1efbe4944ab9f9f7cd02: fix(#8276): retry OIDC token fetch for about a minute (@​fullsend-ai-coder[bot])
• fc7e6f7d060ef8b7598e6d7e99e0614831a35d15: fix(#8276): retry the OIDC fetch up to the 60s window boundary (@​waynesun09)

v0.46.1-rc.1

Changelog

Bug Fixes

• 8ad73777e2d06c62c6bc1efbe4944ab9f9f7cd02: fix(#8276): retry OIDC token fetch for about a minute (@​fullsend-ai-coder[bot])
• fc7e6f7d060ef8b7598e6d7e99e0614831a35d15: fix(#8276): retry the OIDC fetch up to the 60s window boundary (@​waynesun09)

v0.46.0

Highlights

Read this before you upgrade. v0.46.0 removes per-org installation. The per-org commands (github enroll, github sync-scaffold, admin analyze, repos migrate and the rest), the --enroll-all/--enroll-none flags and the mint's ALLOWED_ORGS are gone, and an org-format .fullsend/config.yaml is rejected. Move to per-repo installs first and enroll each repo with fullsend mint enroll <owner/repo>.

repos.yaml must now name an inference method. A repo without inference.auth is no longer treated as Vertex: repos install, repos status and converge fail for it. To keep the old behaviour:

fullsend repos set-default defaults.inference.auth vertex-wif

The repos install Vertex flags are renamed with no aliases: --inference-project, --inference-wif-provider and --inference-region become --vertex-project, --vertex-wif-provider and --vertex-region. On GitLab, CI reads the OpenAI key only from FULLSEND_OPENAI_API_KEY. Run repos install --openai-api-key "$KEY" to set it, then delete the old OPENAI_API_KEY variable.

host_files destinations are stricter. A dest must be an absolute path under /sandbox/ or /tmp/, with no .., and must not name a path the runner owns (for example /sandbox/workspace/.env or /sandbox/workspace/bin/claude). /sandbox/workspace/bin now comes after the image's own directories on PATH, so a host file there adds a command but does not replace one the image provides; to replace an image tool, extend the image. Host scripts no longer inherit JIRA_TOKEN, JIRA_USER_EMAIL or the OTLP header variables unless the harness declares them under env.runner. OTEL_EXPORTER_OTLP_HEADERS must use comma-separated key=value entries (for example Authorization=Bearer%20<token>); a malformed value skips remote export with a warning instead of failing the run. GitLab installs move dispatch to typed pipeline inputs, which needs a short maintenance window: see "Typed-input dispatch migration" in configuring-gitlab.md.

New in this release: OpenAI WIF for repos installs (repos install --inference-auth openai-wif, GitHub only); DCO sign-off with repos install --signoff or signoff: true in repos.yaml; Jira issue links with fullsend issues link --tracker jira --from PROJ-123 --to PROJ-456 --type Blocks, and issues get --fields; and --fullsend-dir now defaults to .fullsend. The sandbox image moves to pi 1.1.0 and Claude Code 2.1.295. Repeated local runs on the same OpenShell gateway no longer fail at profile import, and users with custom GitHub repository roles now get their effective permission.

---

Changelog

Features

• c18e2873d7ee4739cd4fc973657a71ba2f7dbb1b: feat(#7082): add issue type and --fields custom fields to issues get (@​fullsend-ai-coder[bot])
• 6e657f9d9d0f5763fc8b30549b8f54e89a089601: feat(#7084): add fullsend issues link subcommand for typed issue links (@​fullsend-ai-coder[bot])
• 4da4a264f81cc832773d63e31f9f48184071c584: feat(#7304): warn when setup flags pin inherited values (@​fullsend-ai-coder[bot])
• 707b3487a443e480c96988f806d03694ec9cd059: feat(#7771): add GitLab webhook dispatcher CI scaffold (@​fullsend-ai-coder[bot])
• 9dfe094ab20d452a8f68c82934fa8efe1df990f1: feat(#7772): prepare GitLab webhook provisioning and safety checks (@​fullsend-ai-coder[bot])
• 0c1c1f02843aabc4dfe1dd86f27a525434b70fce: feat(#7773): wire gitlab-webhook poll driver to the dispatch spine (@​fullsend-ai-coder[bot])
• 6908711fc6c4e37689a7dad45bee5adf38a619fd: feat(#7979): default --fullsend-dir to .fullsend for agent, run, lock (@​fullsend-ai-coder[bot])
• 94a7d2c233f405306aa88923b27dc0c37dded6f7: feat(#8010)!: require explicit inference.auth selection in repos manifests (@​fullsend-ai-coder[bot])
• 6834eb7f6e2ec3277a6a71fc057a5a474a8cd46f: feat(#8011)!: provision repos inference credentials by selected auth (@​fullsend-ai-coder[bot])
• 1f5e4db43306b6eadeccfaf1bc7be633b71e511e: feat(#8012)!: make repos status and uninstall use selected inference auth (@​fullsend-ai-coder[bot])
• 98fd8d1377aafca5c08ffd579bf6b8e898d93860: feat(#8047): warn when Claude Code reports plugin_errors at startup (@​fullsend-ai-coder[bot])

... (truncated)

Commits

• 3b8fc3f Merge pull request #8283 from fullsend-ai/agent/8282-bump-pi-inference-gateway
• fd52f3b build(#8282): bump pi-inference-gateway to v0.1.1 in the sandbox image
• 6712e62 Merge pull request #8279 from fullsend-ai/agent/8278-gateway-per-model-settings
• ff92c53 docs(#8278): per-model gateway settings with pi-inference-gateway v0.1.1
• 7a93a02 Merge pull request #8275 from fullsend-ai/agent/8274-inference-gateway-docs
• 71b141e Merge pull request #8277 from fullsend-ai/agent/8276-oidc-retry-window
• 831ddbf docs(#8274): say where to run step 5 and narrow the .env note
• fc7e6f7 fix(#8276): retry the OIDC fetch up to the 60s window boundary
• 73b3f7a docs(#8274): fail-closed extension download and absolute harness paths
• 4b0a884 Merge pull request #8271 from fullsend-ai/agent/8262-pi-inference-gateway-ext...
• Additional commits viewable in compare view

[github-actions]

Standardized Dependabot Review Summary

Evidence	Value
Dependency review conclusion	success
Risk	medium
Dependency	fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml
Version	0.45.0
Release age	231 hours
Workflow logs	View run

Eligible for automated approval pending live review-state validation

[github-actions]

Automated dependency approval after static and live review-state validation.
Dependency: fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml@0.45.0
Dependency review result: success
Dependency review conclusion: success
Risk: medium
Release age: 64 hours (minimum: 24 hours)

[github-actions]

Automated dependency approval after static and live review-state validation.
Dependency: fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml@0.45.0
Dependency review result: success
Dependency review conclusion: success
Risk: medium
Release age: 227 hours (minimum: 24 hours)

[github-actions]

Automated dependency approval after static and live review-state validation.
Dependency: fullsend-ai/fullsend/.github/workflows/reusable-prioritize.yml@0.45.0
Dependency review result: success
Dependency review conclusion: success
Risk: medium
Release age: 231 hours (minimum: 24 hours)

[dependabot]

Superseded by #136.