Skip to content

release: v23.5.1 - #1962

Merged
beenuar merged 3 commits into
mainfrom
release/v23.5.1
Nov 18, 2026
Merged

beenuar merged 3 commits into
mainfrom
release/v23.5.1

Conversation

@beenuar

@beenuar beenuar commented Nov 18, 2026 •

Copy link
Copy Markdown
Owner

v23.5.1 — make v23.5.0's release note true

v23.5.0's notes described the "sixteenth secret" — the threat-actor attribution service token — as generated and wired to both ends. On the tag it was neither: ensure_env never minted AISOC_THREATINTEL_SERVICE_TOKEN, and compose delivered it only to the API, not to the threatintel container that enforces the fail-closed gate nor the agents container that sends it. So a fresh make up answered 503 for every /api/v1/actors/* call with every secret apparently in place — a dead feature inside the shipped artifact, and a published claim that wasn't true. This patch fixes that, which is the bar the standing rule sets for a release.

Fixed (3 defects, all found by exercising the shipped tag)

  1. Threat-actor attribution token generated and wired — ensure_env mints it (generated-secret count 15 → 16, reflected in .env.example and the documented count) and compose forwards it to the enforcer and the caller. Pinned by tests/test_threatintel_token_wiring.py, which discovers the readers from source so the generator and the wiring can't drift apart again.
  2. GET /cases?status=open matches the open caseload again — it string-matched the write-path alias literally and returned zero rows, leaving the console's "add to existing case" picker permanently empty. Now expands through case_status.OPEN_STATUSES. Pinned by services/api/tests/test_cases_open_alias.py.
  3. Documentation portal builds again — a generated connector page shipped <your subdomain>.huntress.io, which MDX parses as an unclosed tag, failing every GitHub Pages deploy for six days while per-PR CI stayed green. Generator now escapes it; pinned by tests/test_connector_docs_mdx_safe.py.

The code for all three already merged (#1961, #1960); this PR is the release mechanics only.

Release mechanics

VERSION 23.5.1 (single-source gate green) · CHANGELOG [23.5.1] section dated 2026-11-18 · RELEASES.md What's-new + TL;DR with the 3-defect count · README badge · chart 7.24.1/v23.5.1 (appVersion-only; stable channels hold) · llms.txt/openapi.yaml regenerated from VERSION · changelog-frozen, readme-gates, papers --check, competitor gates green locally.

No breaking changes. Tag v23.5.1 goes on the merge commit; release.yml publishes images; registry verified after.

beenuar and others added 2 commits November 18, 2026 15:44
A patch that makes v23.5.0's own release note true. v23.5.0 described its
'sixteenth secret' — the threat-actor attribution service token — as generated
and wired to both ends; on the tag it was neither, so a fresh make up answered
503 for every /api/v1/actors/* call. ensure_env now mints it (15 -> 16
generated secrets) and compose forwards it to the enforcing threatintel
container and the calling agents container, pinned by a gate that discovers
the readers from source.

Also fixed since v23.5.0: GET /cases?status=open matching zero rows (empty
add-to-case picker), and a six-day docs-portal outage from an MDX-unsafe
generated connector page. README screenshots and the walkthrough video were
re-captured against a live v23.5.0 stack (docs only).

3 defects, all found by exercising the shipped tag. No breaking changes.
Chart publishes as 7.24.1; neither stable channel moves.

VERSION 23.5.1, CHANGELOG [23.5.1] section, RELEASES.md What's-new + TL;DR,
README badge, chart appVersion, llms.txt and openapi.yaml regenerated.
check_release_policy (SLO coverage job) failed on the PR: VERSION read
23.5.1 while apps/web/package.json still read 23.5.0, and the console
footer renders the package version. Same-commit bump, matching how the
v23.5.0 release commit moved it.
check_scoreboard --check (last step of Python — Tests) failed: the newest
substrate row carried v23.5.0 while VERSION reads 23.5.1, and a reader
takes the top row as describing the shipping agent. Refreshed per the
gate's own remedy; the fresh deterministic run re-measured 0.97, matching
the published figure.

commit_sha is stamped as 6887191 (public main) rather than this branch's
local sha: this PR changes no agent code, so the measured agent tree is
identical to that commit, and published rows must name commits that exist
in the public repository.
beenuar added a commit that referenced this pull request Nov 18, 2026
#1964)

## What happened

[PR #1961](#1961) (the visuals
refresh) was pushed from a worktree based on **v23.5.0 (`164b668eb`),
before [#1960](#1960) landed**. The
Git-Data-API replay push serializes the worktree's *entire tree* against
the current parent, so the squash merge silently **reverted all three of
#1960's files**:

- `apps/docs/docs/connectors/huntress.md` → back to the bare `<your
subdomain>` (MDX reads `<your subdomain>` as an unclosed `<your>` JSX
tag);
- `scripts/generate_connector_docs.py` → back to the single-token escape
pattern that can't handle the space in `your subdomain`;
- `tests/test_connector_docs_mdx_safe.py` → **deleted** (it was new in
#1960, absent from the stale tree).

The docs deploy only runs on push to `main` (never per-PR), so #1961's
required checks were all green while this rode in. It went red on the
merge commit (`6887191e7`), re-breaking the portal that #1960 had fixed
for the first time since Nov 12.

## The fix

Restores all three files to their #1960 (`c123ad631`) state. Verified:
- `tests/test_connector_docs_mdx_safe.py` back and green (4 tests);
- a tree-wide scan with `unescaped_jsx_tags()` finds **no** connector
page with an unescaped JSX-like tag;
- `huntress.md:22` is `\`<your subdomain>\`.huntress.io` again.

## Why it matters now

v23.5.1 (in flight as
[#1962](#1962)) documents "the
documentation portal builds again" as a fix — this restores the truth of
that claim before the tag is cut.

## Lesson

The API-replay push captures the worktree's whole tree, so a **stale
base reverts anything merged in between**. Rebase the worktree on
freshly-fetched `origin/main` before pushing. (No code change needed for
the tooling here; this PR fixes the damage.)

Co-authored-by: beenuar <beenuar@users.noreply.github.com>
@beenuar
beenuar merged commit 5f64fb7 into main Nov 18, 2026
149 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant