Skip to content

🌱 sync: top up v6 from v5 (manual conflict resolution, 2026-11-07) - #17024

Merged
clubanderson merged 19 commits into
v6from
sync/v5-to-v6-20261107
Nov 7, 2026
Merged

clubanderson merged 19 commits into
v6from
sync/v5-to-v6-20261107

Conversation

@clubanderson

Copy link
Copy Markdown
Member

Manual v5 β†’ v6 top-up after v6-topup.yml run 42394419186 hit real conflicts.

Carries every v5 commit since the last top-up (incl. #17009 #17011 #16985 #17008 #16983 #16996 #16999 #16992 #16991 #16984 #16785 #16895 #16607, then #17017 #17005 #17019 and the release-stall fix #17015).

Conflict resolutions (v5 behaviour carried into v6 structure):

  • crux/agent/pane_classify.go β€” both guards (lineHasLoginDirective && !authHitIsSourceText || ompLoginScreenLineRe).
  • docs/fleet-report.md β€” v5 rows, path crux/agent/manager.go.
  • docs/operator-reference/README.md β€” v6 rows kept + v5 human_merge_paths text.
  • pkg/github/pr_request_watcher.go, pr_request_retry_reporting_test.go β€” v5 (πŸ› fix(github-relay): report unpushed PR head once, naming the ref checked and the push outcomeΒ #16991 attempts()/attrs); removed dead v6 failures() duplicate from request_retry.go.
  • pkg/scheduler/lane_messages.go β€” v6 variadic refsOnly + block structure, with v5 mergeRow.Reason and spekPublishedContext(issue) line; kickmessage.go keeps both the upstream-sync hint (v6) and spek context (v5).
  • pkg/dashboard/api_governor_security.go, config-watchdog-security.js, static/index.html β€” v6 classifier controls with v5 help text (regenerated, webstatic -check clean).
  • pkg/fleetreport/runtime_test.go β€” v5 shortened message + v6-only ClassAgentPaneTimeout / ClassAgentGitHubCapReached entries.
  • pkg/config/config.go β€” v6 split kept; v5's comment change ported to review_config.go.
  • docs/spektacular.md β€” both sections (v6 Design mode + v5 Publish hand-off); scale-envelope.md citation re-anchored (client.go:5401-5411).
  • .github/scripts/integration-governance.cjs β€” TRUSTED_SCRIPTS_TREE re-pinned to v6 post-merge src/scripts tree 267924f87d77a3fa335accce1b0c74d1a5cfdab8 (governance test 6/6).

Verification: go build ./..., go vet, go test pkg/github, scheduler, fleetreport, config, dashboard all ok; dashboard JS tests 0 fail; docs citations 2007 verified. crux/agent has 10 macOS-only failures (BASH_ENV/read-bit/env-token tests) that fail identically on clean origin/v6 β€” not from this merge.

Merge with a plain merge commit (not squash) so the v5 history is shared.

hivecommons-hive Bot and others added 19 commits November 7, 2026 14:37
The merge-request relay was the last App merge path that ignored
auto_merge.human_merge_pin_only. Wire the policy into the relay's
human-merge-path gate: a PR whose author is in trusted_authors, whose
head equals the requested head (before and after listing files), whose
complete changed-file list classifies as pin replacements, whose every
commit is a verified trusted-author commit, and whose action SHAs are
anchored upstream is admitted without a hold. Anything else keeps the
existing hold, marker comment and refusal; non-GitHub forges stay
fail-closed.

The provenance checks move from the automerge package into
pkg/github/pin_only_provenance.go so the sweep, /approve and the relay
share one implementation. Docs and config comments now state the relay
enforces the exception.

Closes #16265

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…16895)

inspectCanaryEgress drained and closed the request body, then returned
allow when the read failed, forwarding an empty, unscanned body. Fail
closed instead, as enforceLinear and the oversized-body branch already
do, and restore what was read so the denial path sees a well-formed
request.

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…16785)

* fix(crux/proxy): apply destination policy to uninspected tunnels

The plain-HTTP forward path refuses non-routable destinations at dial time,
but a CONNECT to a host the proxy neither inspects nor intercepts, and the
transparent-TLS passthrough for such a host, still dialled whatever host:port
the client named from the proxy's own uid. Route both through the same
guarded dialer so the policy has one definition and one override.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: sec-check <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* test(crux/proxy): pin transparent-TLS refusal counts estate, not loopback

handleTransparentTLS counts a refused estate-bound SNI against the caller
and skips the counter for loopback; no test reached that arm, so dropping
either the reason check or the countViolation call kept the suite green.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: sec-check <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: sec-check <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: sec-check <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ail check (#16984)

* fix(crux/toolapprove): read apply_patch target paths for the guardrail check

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* fix(crux/toolapprove): check every apply_patch path against the guardrails

The guardrail check read only GetFilePath (the first *** header), so a
patch naming a harmless file first and OWNERS second auto-approved.
Iterate GetFilePaths instead and cover it with a Scan-level test.

Hive-Fix-Attempt: 1/6
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…cked and the push outcome (#16991)

The pr-request watcher logged 'head branch is not on the remote β€” the
agent's push did not authenticate' at ERROR on every retry of the same
request (30s→15m backoff), so a single unpushed branch tripped the
fleetreport 5-in-10m threshold and filed a selfreport that asserted an
authentication failure without evidence. The apostrophe in "agent's"
was also treated as an opening quote by fleetreport.NormalizeMessage,
truncating the scope to 'the agentQ'.

- failPRRequest: ERROR once per request with repo, head, the exact ref
  and remote checked, the hive-push-branch relay outcome for that
  branch, and the diagnosis; retries log at Info, and the eventual
  success on retry is logged.
- errMissingHead: describe what to check (relay result.json, credential
  helper, token cache) instead of asserting authentication failed.
- NormalizeMessage: mid-word apostrophes are not quotes.
- hive-open-pr.sh: warn (never refuse) when the head is not on the
  remote before writing the request.

Closes #16969

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
…uired report (#16992)

* πŸ› fix(loginscan): name the backend and login command in the login-required report

The 'login required detected' WARN is what the agent_login_required
fleet selfreport is built from, but it carried only agent, pattern and
sightings. #16953 sat through two decision rounds because
nobody could tell from the report WHICH provider (claude/copilot/gemini/
codex) the supervisor agent on hosted-ibm-alchemy-logg-ptoo had to
re-authenticate.

- Add backend, login_hint (CommandForBackend) and credential_valid to the
  WARN record.
- CommandForBackend: add codex.
- fleet-report.md: document the attrs and the operator action.

Closes #16953

Signed-off-by: Andrew Anderson <andy@clubanderson.com>

* changelog: fragment for #16992

Generated by changelog-fragment-guard from the PR title (#5675).
Edit the wording freely, or replace this file with the no-changelog
label if the change is not user-visible.

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Signed-off-by: clubanderson <407614+clubanderson@users.noreply.github.com>

* changelog: single bullet fragment for #16953

Signed-off-by: Andrew Anderson <andy@clubanderson.com>

---------

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Signed-off-by: clubanderson <407614+clubanderson@users.noreply.github.com>
Co-authored-by: clubanderson <407614+clubanderson@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
…ontributor head (#16999)

A person's Approve stands the hive down from adopting the head, but a red
head cannot be merged as-is either, and a fork author cannot re-run
upstream workflows. The stale-red recovery (update-branch, else re-run
failed jobs) was gated on an in-place adoption, so the flake had no owner
and the hive nudged the contributor to push a new head. Run the recovery
where the admission stand-down happens, and count cancelled workflow runs
(runner stalls) as re-runnable.

Fixes #16998

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
serve.authenticate now attaches a Principal to every request it admits:
the control token (or a token-less server's own unix socket) is the
operator, named by --operator-name / CRUX_OPERATOR_NAME; a lane project
credential is a project principal, accepted only on routes a work-service
family registered (Server.workService) and still refused on agent RPCs.
serve.CallerFrom(ctx) reads it. Prerequisite (S0) for S1 ownership and
S3 gates.

Refs hivecommons/crux#114

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…ce text in agent panes (#17008)

Fixes #17006

Agents that grep or read loginscan/watchdog code paint their pane with the pattern text itself and tripped PaneShowsLogin. Skip grep-prefixed, quoted, alternation and pattern-table lines; the watchdog also only scans the pane tail.

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: clubanderson <407614+clubanderson@users.noreply.github.com>
…ail mode open at L5 (#16985)

* fix(ioscan): apply Input Defense saves without a restart and keep fail mode open at L5

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: williamdalessandro <231250207+williamdalessandro@users.noreply.github.com>

* fix(ioscan): rebuild the classifier on runtime ACMM level changes and fix restart guidance

Review follow-ups for #16985:

- onACMMLevelChanged now calls the new boot.reinstallIoscanClassifier, so
  an L4<->L5 change in the dashboard (or via a hive.yaml reload) applies the
  level's classifier requirement and fail mode without a restart, instead of
  leaving the boot-time classifier and stale alert level in place.
- boot.ioscanAlerts() replaces the two duplicated nil-interface adapters in
  main.go; IoscanConfigApplied is now the shared method value.
- The Classifier backend tooltip in Settings -> Security no longer says
  "Takes effect after a restart"; it says saves there apply immediately and
  a model endpoint or hive.yaml change takes effect after a restart.
- The blocked-kick alert names ioscan.fail_mode (the key the Security tab
  exposes) rather than only ioscan.classifier.fail_mode.
- Docs and changelog fragment updated; tests cover the hook and the alert
  wording.

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* [scanner] fix(dashboard): regenerate static/index.html for the ioscan settings change

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: williamdalessandro <231250207+williamdalessandro@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…cy (#17009)

Hive-Run: #16717
Hive-Plan: contributor-lane-eligibility-parity
Hive-Spec: issue-16717#eligibility-side

Signed-off-by: thatguywob <thatguywob@users.noreply.github.com>
Co-authored-by: thatguywob <thatguywob@users.noreply.github.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Adds the first crux work service from hivecommons/crux#114: a git worktree
pinned at a full 40-hex SHA from an allowlisted repository, owned by the
S0 caller principal.

- checkout/: Service, Config (allowlist, InUse hook, MaxCheckouts), Record,
  typed sentinels; records under <workdir>/checkouts/<id>.json + index.jsonl
- api/checkouts.go: PathCheckouts, CodeTooLarge/CodeUnsupported/CodeUpstream,
  request/response types, client methods
- serve/checkouts.go: POST/GET/DELETE /v1/checkouts[/{id}], ownership via
  principal, error mapping; registered from Handler (families.go is at the
  size ratchet cap) and marked a work service
- cli: crux checkout add|rm|list, local and remote; serve flags
  --checkout-allow/--checkout-dir
- docs/design/crux-work-services.md documents S1 as built, including the
  post-hoc byte cap; TTL sweeper and InUse wiring deferred

Refs hivecommons/crux#114

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…proxy (#17017)

* ✨ crux: let an in-hive Spek publish its RFC and children through the proxy

Allow the GitHub sub-issue and blocked-by dependency writes and the Linear
issueRelationCreate/projectCreate/projectUpdate mutations at ISSUES_ONLY,
add crux/spekmarker to parse the spek:publish provenance marker, and
expose issueshape.SpekPublished as the already-specified signal.

Refs hivecommons/spektacular#114

Signed-off-by: Andrew Anderson <andy@clubanderson.com>

* 🌱 crux/proxy: move hard-deny rules to rules_deny.go to keep rules.go within its size ratchet

Signed-off-by: Andrew Anderson <andy@clubanderson.com>

---------

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…t-fix path with the spek as context (#17019)

Spek (hivecommons/spektacular#109, #114) publishes a finished spec as an
RFC plus child issues into the work source, each body carrying
<!-- spek:publish name=… artifact_id=… target=… [task=…] [rfc=…] -->.

Hive now reads it (pkg/spekmarker): a marker-carrying item is never routed
into a new spec run by runs.triage, is not scored by the clarity gate, and
a child's kick line names the RFC and spek so the agent builds against the
spek's acceptance criteria and writes Fixes #<child>. A marker without
task= is the RFC itself and is treated like epic for
governor.claims.never_claim_labels and IsTrackerIssue, so Hive never claims
the umbrella even in a repo without that label.

Fixes #17018

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Carries the 13 v5 commits (#17009 #17011 #16985 #17008 #16983 #16996 #16999 #16992 #16991 #16984 #16785 #16895 #16607) into v6.

Conflicts resolved: pane_classify.go (both guards), fleet-report.md (v5 rows,
crux/agent path), operator-reference README, pr_request_watcher.go and
pr_request_retry_reporting_test.go (v5 attempts()/attrs version),
request_retry.go (dropped dead v6 failures() duplicate), lane_messages.go
(v6 variadic refsOnly + v5 mergeRow.Reason), api_governor_security.go,
config-watchdog-security.js + static/index.html (v6 classifier controls with
v5 help text), fleetreport/runtime_test.go (v5 message + v6-only entries),
config.go (v6 split kept; v5 comment ported to review_config.go).

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
…gment (#17015)

changelog.d/changed-12923-off-prompt-fixture.md (#12928, 2026-10-19) carried
<!-- release: none -->. compile-changelog.sh folds every fragment into
## Unreleased before derive-release-version.sh runs, so the marker was
present on every decide job since v5.186.4 and each one skipped with
'explicit human override' β€” but the only thing that consumes a fragment is
the release it was forbidding. ~1,600 commits shipped no tag.

- delete the fragment (test-only change; should have been no-changelog)
- check-changelog-fragment.sh / compile-changelog.sh reject release:none in
  a fragment; major|minor|patch bump markers remain legal
- derive-release-version.sh emits a ::warning:: when a hand-placed none
  marker freezes releases, and fails the decide job once the freeze has
  outlived RELEASE_FREEZE_MAX_DAYS (default 7) since the newest tag
- derive-release-version.sh: the Added-heading inference used grep -q under
  pipefail; on the 558 KiB compiled backlog printf took SIGPIPE and the
  pipeline failed => silent patch instead of minor. grep now reads to EOF.
- docs: changelog.d/README.md, src/docs/releases.md
- tests: test-compile-changelog.sh 5b/5c, test-derive-release-version.sh
  3b (large body) and 5b (freeze age)

Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Signed-off-by: Andrew Anderson <andy@clubanderson.com>

# Conflicts:
#	.github/scripts/integration-governance.cjs
#	src/docs/scale-envelope.md
#	src/docs/spektacular.md
#	src/pkg/scheduler/kickmessage.go
#	src/pkg/scheduler/lane_messages.go
@kubestellar-prow kubestellar-prow Bot added the dco-signoff: yes Indicates the PR's author has signed the DCO. label Nov 7, 2026
@kubestellar-prow

Copy link
Copy Markdown
Contributor

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign danathar for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@kubestellar-prow kubestellar-prow Bot added the size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. label Nov 7, 2026
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Open PR overlap check

Base branch: v6

Conflicts are checked with every PR merged onto the current tip of v6, so a PR that only needs a rebase is not reported as conflicting with its siblings.

No conflicting open PRs or same-file overlaps were found.

@hivecommons-hive

Copy link
Copy Markdown
Contributor
2c1baa5

ℹ️ Sentinel notice β€” informational; author is trusted (owner), merge is not blocked

Hive flagged this PR (author @clubanderson, head 2c1baa541610) because it matches behaviors that can override security controls, escalate privileges or damage the codebase. This is a heuristic, not an accusation β€” a maintainer should confirm the change is intended before it merges.

  • sensitive_path β€” changes 3 sensitive path(s) (Touches a sensitive path (OWNERS, workflows, policies, hive config, security docs, …))
    • src/crux/proxy/rules.go
    • src/crux/proxy/rules_deny.go
    • src/crux/proxy/rules_test.go

Hive did not add the sentinel-alert label because this author is trusted. The finding remains recorded for audit and the dashboard Security tab; set sentinel.trusted_authors_block: true to require the blocking label for trusted authors too.

@clubanderson
clubanderson merged commit 7b9eeb8 into v6 Nov 7, 2026
88 of 94 checks passed
@kubestellar-prow
kubestellar-prow Bot deleted the sync/v5-to-v6-20261107 branch November 7, 2026 17:49
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Thank you for your contribution! Your PR has been merged.

We'd love to hear how your experience was: share feedback

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dco-signoff: yes Indicates the PR's author has signed the DCO. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant