work-items(triage): "raw intake = items the team did not author" scope prose excludes team-authored self-observation/dogfood issues that legitimately need triage

[kyle-sexton]

Lane: triage · Category: behavioral (scope-definition gap)

Triggering example

2026-07-19 triage loop cycle. The needs-triage queue held 7 issues (#459, #475, #477, #478, #479, #480, #481), all authored by the team's own members and filed by the skill's own self-observation / dogfood rule ("file every distinct problem/improvement you notice as an issue … label priority: needs-triage"). They carried only priority: needs-triage, so they correctly surfaced in the attention-view queue and genuinely needed triage (priority normalization, tier routing, plan drafting).

But the triage SKILL.md "Scope: raw intake only" section defines the flow's input as:

raw intake — items the team did not author (bug reports, incoming feature requests, unsolicited PRs)

By that prose, team-authored dogfood issues are out of scope — yet they are exactly what the queue contains and what the lane exists to triage.

Observed vs expected

• Observed: two skill statements are in tension. (1) Scope says raw intake = "items the team did not author." (2) The same plugin ships a self-observation rule that has the team author priority: needs-triage issues, plus an attention view that surfaces every priority: needs-triage item regardless of author. An operator reading the scope prose literally would exclude the very items the queue is built from.
• Expected: the scope definition keys on triage state, not authorship — an item is raw intake if it is untriaged (unlabeled or status: needs-triage / priority: needs-triage), whoever authored it. The "did not author" phrasing should become an illustrative list of common raw-intake sources, not the defining boundary. The real exclusion the section wants is "never re-triage already-triaged output" (decompose / track add items born with classification labels), which is authorship-independent and already stated separately.

Category

bug (behavioral/definitional) — the scope prose contradicts the plugin's own self-observation rule and attention-view construction.

Note

Distinct from #459 (wait-gate vs autonomous mutation), #478 (routing-rule absorption), and #449 (no-local-checkout mode). This is specifically the authorship-based scope boundary excluding legitimately-queued team-authored intake.

[kyle-sexton]

This was generated by AI during triage. · Category: behavioral (definitional) · Priority: priority: high (pipeline correctness — triage skill). Routed status: ready.

Plan (T1 — single surface: the "Scope: raw intake only" section)

Re-key the scope definition on triage state, not authorship. An item is raw intake if it is untriaged (unlabeled or carrying the raw marker), whoever authored it. Demote the "items the team did not author" phrasing to an illustrative list of common sources, not the defining boundary. Keep the real exclusion — "never re-triage already-triaged output (decompose / track add items born with classification labels)" — which is authorship-independent and already stated separately.

Acceptance

• The scope section admits team-authored self-observation/dogfood issues carrying the raw marker as valid raw intake.
• The decompose/track add "born triaged" exclusion is preserved and stated on a state basis, not authorship.
• No contradiction remains between the scope prose, the self-observation filing rule, and the attention-view construction.

[kyle-sexton]