In this article
backlog-plan
| Field | Value |
|---|---|
| Kind | skill |
| Source | .github/skills/project-planning/backlog-plan |
| Invocation | Invoked directly as /backlog-plan, or loaded on demand by referencing agents |
| Interactive | No |
What it does
Read-only backlog planning for Azure DevOps, GitHub, and Jira. Use to discover, triage, sprint-plan, or resume without mutating a tracker.
When to use it
Use this skill to discover candidate work, retrieve assigned items, prepare an implementation handoff, triage issues, plan a delivery window, or resume planning. It reads one resolved tracker and writes local planning artifacts, but never creates, updates, comments on, or closes tracker items.
Choose functional-planner for PRD-to-hierarchy conversion and backlog-execute
for approved tracker changes. Read access and the shared backlog-management
contract must be available; failure does not justify choosing another tracker.
Example usage
Ask: /backlog-plan triage open documentation issues in example-org/sample-service on GitHub; recommend priorities and duplicates without changing anything. Use
your actual authorized repository in place of the example. Expect hydrated issue
evidence, similarity assessments, a planning log, and a handoff of proposed actions.
Success means recommendations are tied to acceptance criteria and ambiguous
duplicates remain questions for human review.
For a delivery window, use sprint with its milestone or iteration and supplied
capacity to get coverage, dependency, and gap analysis rather than item mutations.
Use my-work to retrieve assigned items, then task-plan to enrich selected work
for implementation. resume reads durable planning records and proposes the next
step; missing logs or unresolved identifiers are blockers, not permission to
recreate work. Applying any proposed changes requires a separate execution pass.