Skip to content

mk:jira-evaluator

What This Skill Does

Forks the jira-evaluator agent to analyze a single Jira ticket for complexity and inconsistencies. Read-only — never mutates Jira. Persists the evaluation to tasks/reports/jira-evaluate-*.md for cross-session continuity.

When to Use

  • Triggers: "evaluate PROJ-123", "analyze ticket complexity for KEY", "check ticket quality"
  • NOT for: estimation (mk:jira-estimator) · full RCA (mk:jira-analyst).

Output

markdown
## Ticket Evaluation: {ISSUE-KEY}

**Complexity:** {Simple|Medium|Complex} (likely {Fibonacci range}pt)
**Confidence:** {High|Medium|Low}

### Signals
- Scope, Dependencies, Regression risk, Requirement clarity, External, Historical
- Workflow shape (read from tasks/jira-workflows/<workflow-slug>.md)

### Issues Detected
- ⚠️ {issue}

### Suggested Actions
- {actionable suggestions referencing the right mk:jira-* leaf}

### Open Questions
- {unresolved ambiguities}

Workflow Cache Integration

The evaluator reads the discovered workflow cache (tasks/jira-workflows/<workflow-slug>.md) — workflow shape (status_count, transition_count, parallel branches) feeds the complexity signal. If the cache is absent, the agent runs fetch-workflow.sh <KEY> first.

Report Persistence

Writes to tasks/reports/jira-evaluate-{YYMMDD}-{HHMM}-{ISSUE-KEY}.md. Consumed by:

  • mk:jira-estimator — auto-incorporates complexity signals
  • mk:planning-engine — capacity analysis
  • mk:cook (plan-creation input)

Shared References

  • references/evaluation-rubric.md (in mk:jira/references/)

Peer Skills

mk:jira-estimator (run after evaluator for point estimate) · mk:jira-analyst (full ticket synthesis) · mk:planning-engine (downstream consumer)

Agent

jira-evaluator — A + C (FS write only — NEVER Jira state), NOT B. 2/3 compliant.

Released under the MIT License.