[Tech Debt] Reconcile the GitHub label set with the vendored vocabulary the issue templates require

[chris-prener]

Parent epic: #123 — [Epic K] Vendored Workflow Toolkit Adoption & Governance

User story

As a contributor filing an issue in this repo, I want the labels the issue templates ask for to actually exist, so that I can complete the Definition of Ready instead of being asked for a label the repo cannot apply.

Motivation

The vendored toolkit's LABELS.md describes a canonical vocabulary of 35 labels. This repo's live GitHub label set has 18. The gap is not cosmetic — several vendored workflows read labels as control flow:

• pr_orchestrator/SKILL.md Step 1 resolves a gate profile from a gate-profile/{light,standard,full} label on each closing issue. No such label exists here, so every issue silently resolves to standard. That happens to be the sane default, but it is a default reached by absence rather than by decision, and the light and full tiers are unreachable.
• next_action/SKILL.md and backlog_grooming/SKILL.md rank by scoring bucket.
• The bug_report, feature_request, tech_debt, and qc_finding templates each instruct the filer to apply "exactly one of" a scoring bucket and "exactly one of" a priority including priority/blocker.

Current pain

Concretely missing from this repo:

Family	Vendored vocabulary	Present here
Scoring	wsjf/critical|high|medium|low (this repo's pinned LABELS.md) — or tier/quick-win|big-bet|fill-in|reconsider (upstream default per ADR-0014)	neither
Priority	priority/blocker | high | medium	priority/high, priority/medium only
Gate profile	gate-profile/light | standard | full	none

Two knock-on effects:

1. The templates are unsatisfiable. A filer following tech_debt.md is told to apply a wsjf/* bucket and offered priority/blocker. Neither can be selected. The DoR mini-checklist line "Labels applied (Type + Priority + WSJF bucket)" can never be ticked honestly. Note also that non-collaborators cannot set labels at all on a public repo, so for external contributors the instruction is doubly inert.
2. .github/labels.json is not the source of truth and is not reconciled. It is tracked here, LABELS.md is pinned via local_override: true in the vendor manifest, and nothing keeps the two — or either one and the live repo state — in agreement. Upstream sync only moves files; labels are repo state, so no sync operation will ever close this gap. (Filed upstream as pfizer-evgen/agentic-dev#398.)

Proposal

Decide the target vocabulary first, then make repo state match it. Three viable shapes, in rough order of preference:

A. Adopt a trimmed subset (recommended). Keep this repo's wsjf/* choice, create the four wsjf/* labels plus priority/blocker, and explicitly decline gate-profile/* — documenting in LABELS.md that all issues run the standard profile. Smallest change that makes the templates truthful.

B. Adopt the full 35. Create everything LABELS.md names. Maximum toolkit fidelity, but imports vocabulary this repo has no demonstrated use for, on a public repo where the label sidebar is contributor-facing.

C. Trim the templates instead. Strip the scoring block and the priority/blocker mention from the four issue templates and shrink LABELS.md to the 18 that exist. Lowest ceremony; gives up WSJF/tier ranking entirely.

Whichever is chosen, .github/labels.json should become the machine-readable source and a small reconciliation step (gh label create / gh label edit driven from that file) should exist so drift is detectable rather than discovered during a sync.

Decide before implementing — this is a vocabulary decision, not a mechanical one, and the wrong choice is annoying to reverse once labels are applied to real issues.

Acceptance criteria

• A decision is recorded (comment or ADR) on which of A/B/C this repo adopts, and on whether gate-profile/* is in or out.
• The live GitHub label set matches the chosen vocabulary.
• .github/labels.json and .github/LABELS.md agree with each other and with live repo state.
• The four issue templates reference only labels that exist.
• The DoR mini-checklist in each template is satisfiable as written.
• LABELS.md notes that it is pinned via local_override and why, so a future sync does not silently re-diverge.

Out of scope

• Retroactively labelling existing open issues with new scoring buckets.
• Adopting the upstream tier/* vocabulary in place of wsjf/* — that is a separate decision, deliberately deferred when the templates were pinned in PR chore: sync vendored agent toolkit to renamed upstream agentic-dev #117.
• Any change to the gate-profile mechanism in pr_orchestrator/SKILL.md (vendored; not ours to edit).

Codebase context

• Affected files: .github/labels.json, .github/LABELS.md, .github/ISSUE_TEMPLATE/{bug_report,feature_request,tech_debt,qc_finding}.md.
• Origin: PR chore: sync vendored agent toolkit to renamed upstream agentic-dev #117 Notes, follow-up 1 of 2. Surfaced during the vendored-toolkit sync ([Tech Debt] Sync vendored agent toolkit to renamed upstream pfizer-evgen/agentic-dev #116).
• Upstream counterpart: pfizer-evgen/agentic-dev#398 (labels are repo state; sync only moves files).
• Related: the four templates and LABELS.md are pinned local_override: true in .github/VENDOR_MANIFEST.json, so upstream will not overwrite whatever is decided here.

Definition of Ready (mini-checklist)

• Motivation + current pain are concrete

• Acceptance criteria measurable

• Out of scope explicit

• Labels applied (Type + Priority + WSJF bucket) — WSJF bucket label does not exist; that is the subject of this issue

• Parent epic linked above (standalone)

• Sibling follow-up: [Tech Debt] Tracked public docs link into gitignored .github/vendored-decisions/, so those links 404 #119 (tracked docs linking into gitignored .github/vendored-decisions/).

[chris-prener]

Triage — 2026-08-10

• Route: Route 1 — re-parented from standalone to [Epic K] Vendored Workflow Toolkit Adoption & Governance #123 ([Epic K] Vendored Workflow Toolkit Adoption & Governance) via native sub-issue link.
• Assessment: DoR met — three concrete options with a stated recommendation, measurable acceptance criteria, explicit out-of-scope. The original standalone justification correctly ruled out Epics H/I/J; Epic K did not exist at filing time. Labels unchanged (tech-debt, priority/medium).
• Sequencing: blocked by [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120. This issue's AC1 asks "whether gate-profile/* is in or out" — that is the same decision [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120 owns and hands to the Tech Lead. Starting here first would either duplicate that decision or hard-code an answer [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120 subsequently overturns. Recommend picking this up only after [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120 records its adopt-or-decline outcome, at which point option A/B/C here becomes largely mechanical.
• Note on option A: the recommendation to "explicitly decline gate-profile/*" is consistent with what [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120 may decide, but is not this issue's call to make. Treat the decline as an input from [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120, not an assumption here.

No repo files were modified. Body edit was limited to the **Parent epic:** line.

---

_{Co-authored-by: Copilot}

[chris-prener]

Implementation plan

Last updated: 2026-08-11T13:53:41Z by Developer
Status: drafting
Sync token: 5254189896:2026-08-11T14:00:52Z

Approach

This issue's own body requires a target label-vocabulary decision (A/B/C) to be made and recorded before any implementation work is scoped: option A (adopt a trimmed subset — wsjf/* plus priority/blocker, decline gate-profile/*), B (adopt the full 35-label vendored vocabulary), or C (trim the issue templates and LABELS.md down to the 18 labels that already exist). The issue explicitly says this is "a vocabulary decision, not a mechanical one, and the wrong choice is annoying to reverse once labels are applied to real issues" — out of Developer-persona scope. The plan for this issue is therefore to hand off to Tech Lead to make and record the decision; once it lands, this plan will be updated (via Update) with the concrete label-creation / template-edit steps that follow from the chosen option.

Files to touch

• None yet — no implementation files identified until Tech Lead records the vocabulary decision.

Verification

• Tech Lead decision recorded (comment or ADR) on A/B/C and on whether gate-profile/* is in or out
• Remaining verification steps TBD once the decision is recorded and this plan is updated (live label set, .github/labels.json, LABELS.md, and issue templates all need to agree per the issue's acceptance criteria)

Risks / open questions

• Blocking: this decision interacts with [Tech Debt] Resolve ADR-0014 numbering collision and adopt/decline gate-profile labels #120 (gate-profile ADR) — the gate-profile/* label question is shared between both issues and should be resolved consistently.
• .github/labels.json is not currently the reconciled source of truth; the eventual implementation will need a small reconciliation step so drift is detectable going forward, per the issue's proposal.

Where I left off

Not yet started — awaiting Tech Lead decision.

Decisions made

• 2026-08-11: authored against commit e0875c2
• 2026-08-11: identified that this issue requires a vocabulary/governance decision out of Developer scope; routed to Tech Lead rather than proceeding on an assumed option