Problem
Issues are created ad-hoc during interactive sessions with inconsistent structure. Some have `## Dependencies`, some don't. Some have `## Acceptance Criteria`, some use `## Testing Strategy`. The triage/harden pipeline then spends time restructuring them. A consistent template at creation time would reduce downstream friction.
Solution
Create an implicit skill (`.claude/skills/issue-template/SKILL.md`) that auto-activates when the user is creating a GitHub or Jira issue. The skill provides the canonical issue body structure that works best with SOVA's pipeline:
- `## Objective` (required): what and why, 1-3 sentences
- `## Acceptance Criteria` (required): checkbox list, testable outcomes
- `## Dependencies` (optional): `- #NNN` or `- Part of epic #NNN` format (parsed by `dependency_graph.py`)
- `## Context` (optional): relevant background, links, prior art
- `## Scope Boundaries` (optional): explicit out-of-scope items
The skill should be:
- Implicit (skill, not command): auto-activates when the context matches issue creation. No need to invoke `/create-issue`.
- Tracker-aware: adapts field names for Jira (e.g., maps Acceptance Criteria to Jira's standard fields).
- Distributable: ships with `sova install` to external projects via `skills/` directory.
- Minimal: provides the template and constraints, not a multi-step workflow.
Design Decisions
- Skill over command? Yes. The user said "implicit" is better. The skill auto-fires when Claude detects issue creation intent (filing a bug, proposing a feature, creating a task). A command would require the user to remember to invoke it.
- Where does it live? `skills/issue-template/SKILL.md` (distributable) + `.claude/skills/issue-template/SKILL.md` (SOVA-specific copy with SOVA-specific fields like `area:` label suggestions).
- Triage quality gate alignment: the skill template must match the fields checked by `TriageRole.compute_quality_score()` so issues created with the template always pass the quality gate.
Dependencies
None.
Problem
Issues are created ad-hoc during interactive sessions with inconsistent structure. Some have `## Dependencies`, some don't. Some have `## Acceptance Criteria`, some use `## Testing Strategy`. The triage/harden pipeline then spends time restructuring them. A consistent template at creation time would reduce downstream friction.
Solution
Create an implicit skill (`.claude/skills/issue-template/SKILL.md`) that auto-activates when the user is creating a GitHub or Jira issue. The skill provides the canonical issue body structure that works best with SOVA's pipeline:
The skill should be:
Design Decisions
Dependencies
None.