Improvement Backlog — HELIX 2026-Q2

Source identity (from 06-iterate/improvement-backlog.md):

ddx:
  id: improvement-backlog
  depends_on:
    - metrics-dashboard

Improvement Backlog — HELIX 2026-Q2

Iteration: 2026-Q2 (post-v0.3.3) Source Learnings: docs/helix/06-iterate/alignment-reviews/AR-2026-05-01-public-examples.md, docs/helix/06-iterate/metrics-dashboard.md, docs/helix/06-iterate/security-metrics.md, recent tests/helix-cli.sh regression history, the active concerns roster (hugo-hextra, demo-asciinema, e2e-playwright).

Prioritization Rules

Rank by authority leverage: items that shore up CONTRACT-001 alignment or fix authority-inversion risks rank above cosmetic improvements.
Rank by public-surface impact: items that make the public website / artifact reference more accurate to readers rank above purely-internal cleanups.
Items without tracker references stay at the bottom until they are filed — the backlog does not retain unsourced ideas.
Safety / supply-chain risk items take priority over feature work. None are currently flagged in security-metrics.md.

Backlog Items

Priority	Item	Evidence	Tracker Ref	Why Now	Status
P1	Author HELIX-real worked examples for the 12 descoped artifact slugs	AR-2026-05-01-public-examples.md “Follow-up beads to file” table	This iterate cycle is consuming them; one bead per slug	Public artifact reference currently shows a `Decision / Why It Exists` page where a worked example is expected, misleading readers	open
P1	Restore the public-site Playwright proof lane to green	`e2e-playwright` concern; recent CI history	helix-438c8a07 (and successors under `area:site,area:testing`)	Site regressions are landing without test coverage; Playwright is the only deterministic site verification	open
P2	Reconcile `helix-run` ↔ `helix run` cosmetic drift across PRD, ADR-001, SD-001, TD-002	AR-2026-05-01-public-examples.md cross-cutting findings #7	Filed as part of artifact-restoration work; covered by per-doc refresh beads	Published canonical examples should use the actual CLI invocation form; cosmetic but visible	open
P2	Verify the `github.com/easel/ddx` repo URL in architecture.md vs the canonical org URL	AR-2026-05-01-public-examples.md cross-cutting findings #6	Pending bead under `area:docs,area:artifacts`	Public links must resolve; org migration may have invalidated this link	open
P2	Propagate context-digest concerns onto open beads under FEAT-006	FEAT-006 status in implementation-plan-2026-04-11-snapshot (“DIVERGENT”)	helix-674b1b42, helix-691d18c0, helix-d9f93a59	Concern propagation is incomplete; beads claimed today may not see the right practices	open
P3	Replace remaining hardcoded `/home/erik/...` paths in older docs	AR-2026-05-01-public-examples.md cross-cutting findings #4	One bead per affected doc	Public-readability blocker for any doc flipped into `HELIX_REAL_EXAMPLES_PUBLISHABLE` later	open
P3	Demo-recording inventory refresh under `demo-asciinema`	`demo-asciinema` concern; demos may have drifted from current CLI surface	helix-39fc1526 and successors	Public demos should match shipped behavior post-`v0.3.3`	open
P3	Forge-agent discovery scaffolding	`eff9742` “chore: add forge agent discovery tracker issue”	The bead created in that commit	Listed as a known known-issue in `release-notes.md`; no current path to consume it	open

Selection for Next Iteration

Chosen item: P1 — Author HELIX-real worked examples for the 12 descoped artifact slugs. This iterate cycle is consuming them as the current work-unit; eight slugs are flipped on as of v0.3.3, and twelve more are pending per AR-2026-05-01.
Why it wins the next slot: Highest authority leverage (CONTRACT-001 alignment), highest public-surface impact (the artifact reference is the most-visited part of the site), and it unblocks all P3 items below by exercising the relocate-meta-decision-doc pattern that several other cleanups will reuse.

Review Checklist

Each item cites evidence
Tracker references are included (or marked as “pending bead” where not yet filed)
Ordering is deterministic (priority + position within priority)