feat(cymbal): set posthog-elixir wire-order cutoff at 2.13.0

[cat-ph]

Problem

Seventh SDK flip of the wire-order standardization (PostHog/sdk-specs#11): posthog-elixir ships canonical bottom-up frames in 2.13.0 (PostHog/posthog-elixir#160, latest release 2.12.1 + that PR's lone minor changeset; no other changesets pending).

This one was deliberately held until the elixir grouping fix (PostHog/posthog-elixir#164, released in 2.12.0) proved out: fleets on 2.12.x now fragment at ~2.4% unique fingerprints per event versus 40–99% on older versions, so the flip lands on a healthy grouping baseline.

Changes

• posthog-elixir gets canonical_since: Some(2.13.0); everything below (or unparseable) keeps normalizing.
• Mirror of the established cutoff gate test: 2.12.1 / 2.1.0 / garbage / missing normalize; 2.13.x, 2.14.0, 3.0.0 pass through; legacy-order reconstruction ignores the cutoff.

Merge sequencing

1. Re-verify posthog-elixir's latest release is still 2.12.x immediately before merging (the standing version-claim rule).
2. Merge this, wait for cymbal autodeploy.
3. Merge feat(error-tracking): emit stack frames in canonical bottom-up order posthog-elixir#160 and release as 2.13.0; no other elixir minor in between.

Watch et-wire-order-launch as with the previous six. Elixir frames are platform: "custom" (no resolution reshaping), so post-cutoff legacy reconstruction is byte-exact — no inline-expansion caveat here.

How did you test this code?

Automated only: new elixir_cutoff_gates_normalization_by_version unit test; normalization module green (15 tests); clippy -D warnings and fmt clean. Seventh instance of the reviewed cutoff pattern. The companion flip branch was rebased onto elixir main v2.12.1 (which includes #164's handler changes — verified it introduces no additional frame-building path) with the full elixir suite green (322 passed) on the pinned toolchain.

Automatic notifications

• Publish to changelog?
• Alert Sales and Marketing teams?

Docs update

Rollout process documented in the sdk-specs change.

🤖 Agent context

Autonomy: Human-driven (agent-assisted)

Claude Code authored this under my direction as step 7 of the flip rollout, after validating in production data that the #164 grouping fix works for upgraded fleets (2.4% unique fingerprints on 2.12.0 vs 91–99% on 2.4–2.5). Target version derived from the latest release plus #160's pending changeset.

[greptile-apps]

_{Reviews (1): Last reviewed commit: "feat(cymbal): set posthog-elixir wire-or..." | Re-trigger Greptile}

[stamphog]

Small, focused addition of one SDK version-cutoff entry mirroring an already-reviewed pattern (six prior instances), with a matching unit test; author is on the owning error-tracking team, which supplies the required independent assurance for this ingestion-path change, and there are no unresolved concerns or holds.

• Author wrote 0% of the modified lines and has 16 merged PRs in these paths (familiarity MODERATE).
• 👍 on the PR from chatgpt-codex-connector[bot], greptile-apps[bot], hex-security-app[bot].

Gate mechanics and policy version

Gate		Result
prerequisites	✓	all clear
deny-list	✓	no deny categories matched
size	✓	28L, 1F substantive — within ceiling
tier	✓	T1-agent / T1b-small (28L, 1F, single-area, feat)
stamphog 2.0.0b3		.stamphog/policy.yml @ 48d972a · reviewed head b9a2c04

[deployment-status-posthog]

Deploy status

Environment	Status	Deployed At	Workflow
dev	✅ Deployed	2026-07-23 19:15 UTC	Run
prod-us	✅ Deployed	2026-07-23 19:41 UTC	Run
prod-eu	✅ Deployed	2026-07-23 19:47 UTC	Run