SuperAPI AI Directory

issue-todo

Coding & development

View repository

Updated 10-15 22:09

About this skill

script path

Among the following, the first path that exists를 <skill> is used.

.claude/skills/issue-todo      # 현재 프로젝트 (Claude Code)
.codex/skills/issue-todo       # 현재 프로젝트 (Codex)
~/.claude/skills/issue-todo    # 홈 설치
~/.codex/skills/issue-todo     # 홈 설치
Execution order

Step 0 — Verify prerequisites

git rev-parse --show-toplevel

Tracker authentication is verified by the script via issue-tracker.mjs. For github authentication failure, gh-setup finish it first with the skill.

Step 1 — Update the graph (sync)

node <skill>/scripts/issue-todo.mjs sync

Update the tracker's open and closed issues as nodes, and automatically detect references like "depends on #N" in the body as edges. If the output's CYCLE= is not empty, there is a cycle, so check with validate . For details, references/dag-ops.md.

Step 2 — Reinforce dependencies (link, optional)

Manually add dependencies that are not caught by automatic detection, along with their rationale.

node <skill>/scripts/issue-todo.mjs link <from> <to> --type depends-on --why "<근거>"

Edges that create a cycle are rejected. Remove an incorrectly added edge with unlink <from> <to> .

Step 3 — Produce the todo (plan)

node <skill>/scripts/issue-todo.mjs plan

Using topological sort + ready-frontier + priority, actionable / in progress / blocked / done are classified. --json as machine output. Without arguments, issue-todo if you call only this, this step is produced.

Step 4 — Recommend the next item to start (next)

node <skill>/scripts/issue-todo.mjs next

From the ready-frontier, pick the first issue by priority and number, and /issue-start #N suggest it.

Step 5 — Check (validate)

node <skill>/scripts/issue-todo.mjs validate

Check for cycles (exit 2 if any), dangling edges, and close mismatches.

Final report

그래프    .issue/graph.json (노드 <n>개 / 엣지 <m>개)
착수 가능  #<..> (ready-frontier)
막힘      #<..> ← 대기 #<..>
진행 중    #<..>
점검      사이클 <유/무> / dangling <n> / 불일치 <n>
다음      <사용자가 고른 행동>