| name | 3a-track |
| description | Use when a developer wants to check progress, understand where they are in their agent project, or figure out what to do next. Reads the roadmap and project state, then provides a status report with next steps. Triggers on: "where am I", "what's next", "status", "progress", "what should I do now", "3a status", "track", "stuck", "lost", "which phase", "next step". Not for creating a plan — use 3a-plan first. Not for stage-gate reviews — use 3a-review.
|
| allowed-tools | Read Write Grep Glob Bash |
| metadata | {"type":"skill","version":"0.1.0"} |
track
Show where the developer is in their agent project and what to do next.
When to use
- Developer returns to a project after a break and needs orientation
- Developer finished a task and wants to know what's next
- Developer feels stuck or lost
- Developer wants a progress overview
Do NOT use for:
- Creating a plan from scratch -> use
3a-plan
- Formal readiness checks before deploy/production -> use
3a-review
- Running
agentcore commands -> delegate to aws-agents
Input
$ARGUMENTS can be:
- Empty — full status report
- "next" — just the next action
- "phase" — current phase details
- "blockers" — what's preventing progress
Process
Step 0: Check prerequisites and read project state
Check that aws-agents plugin is installed. This plugin delegates implementation guidance to aws-agents skills. If aws-agents is not available, stop and tell the developer:
3A Plugin은 aws-agents 플러그인과 함께 사용해야 합니다.
설치 방법: claude plugin add --from https://github.com/aws/agent-toolkit-for-aws/tree/main/plugins/aws-agents
설치 후 다시 실행해 주세요.
Read project state. Read these files to understand the project:
.3a/roadmap.md — the execution plan
.3a/architecture.md — architecture decisions
.3a/requirements.md — what's being built
agentcore/agentcore.json — current project config (if exists)
If .3a/ doesn't exist, tell the developer:
계획이 아직 없습니다. /3a-plan을 먼저 실행해서 로드맵을 만들거나, 프로젝트에 대해 알려주시면 방향을 잡아드리겠습니다.
Step 1: Detect current phase
Check project state signals to determine which phase the developer is in:
| Signal | Phase |
|---|
No agentcore/ directory | Phase 1: Scaffold (not started) |
agentcore/agentcore.json exists but no custom tools/logic | Phase 1: Scaffold (in progress) |
| Agent code has custom tools/system prompt | Phase 2: Implement |
agentcore status shows deployed resources | Phase 3: Deploy |
| Monitoring/guardrails configured | Phase 4: Harden |
Run these checks:
ls agentcore/agentcore.json 2>/dev/null
find . -name "main.py" -path "*/app/*" 2>/dev/null | head -1
agentcore status 2>/dev/null | head -20
Step 2: Compare against roadmap
Read .3a/roadmap.md and match the current state to roadmap tasks:
For each task in the current phase:
- Done: evidence found in project files
- In progress: partially complete
- Not started: no evidence found
Step 3: Produce status report
Format the report as:
## 현재 상태
**단계**: [N] - [단계명]
**진행률**: [X]/[Y] 작업 완료
### 완료됨
- [작업] — [근거]
### 진행 중
- [작업] — [완료된 부분, 남은 부분]
### 미착수
- [작업]
## 다음 할 일
[지금 가장 중요한 한 가지]
**방법**: `[aws-agents 스킬]`을 사용하여 [구체적 작업].
## 차단 요소 (있는 경우)
- [차단 요소] — [해결 방안]
Step 4: Update roadmap
If the status check revealed completed tasks that weren't marked done, update .3a/roadmap.md with checkmarks:
- [x] Create project with agentcore create
- [x] Verify local dev works
- [ ] Add custom tools <-- current
- [ ] Connect external API
Step 5: Proactive guidance
Based on the current state, proactively suggest:
- If stuck on a task: explain what the task involves and which
aws-agents skill helps
- If between phases: suggest running
3a-review before moving to the next phase
- If off-roadmap work detected: ask if the roadmap needs updating
- If no activity for a while: ask what's blocking progress
Always end with a concrete next action, not a generic list.
Output
- Status report with current phase, progress, and next action
- Updated
.3a/roadmap.md with progress markers
- Specific guidance on what to do next and which tool to use
Quality criteria
- Status detection is based on actual project state, not assumptions
- Next action is specific and actionable, not generic advice
- Each recommendation maps to a concrete
aws-agents skill or command
- The developer knows exactly what to do when the report is done