standup

name	standup
description	Use when a team exists but current status is unknown — backlog changed, context switch happened, or session just started. Syncs status, surfaces blockers, checks learning health. Keywords: status, sync, blockers, progress, team, check-in, reorient.
argument-hint	[focus area]
disable-model-invocation	false
user-invocable	true
allowed-tools	Read, Grep, Glob, Bash(git:*)

Standup: Team Status Sync

You are running a Standup -- a quick, read-only status check with a persistent learning team. Each member's status is derived from their learnings file, git activity on owned files, and the task backlog.

Focus area (optional): $ARGUMENTS

When to Use

• Team manifest exists but you don't know who is active, blocked, or ready
• Task backlog has changed since the last standup (new tasks, closed tasks, status shifts)
• Re-entering a session after a context switch and need to reorient before deciding what to dispatch
• Before planning a sprint, to confirm team state rather than assume it

Don't Use When

• No .claude/team.yaml exists — run /assemble first
• You just ran standup in this session and nothing has changed (git log and backlog unchanged)
• You need to assign or dispatch work — standup is read-only; use /sprint to dispatch
• You need a solo status check with no team — use /status instead

Overview

Validate preconditions
  -> Load team manifest + task state + git log
    -> Gather per-member activity (git + learnings)
      -> Verify data completeness
        -> Synthesize board

---

Phase 1: Load Context

1a. Read Team Manifest

Read .claude/team.yaml to discover members, their roles, and ownership patterns.

If no .claude/team.yaml exists, stop here: "No team manifest found. Run /assemble to create one."

Note the member count. You must produce a status entry for every member — missing members make the standup incomplete.

1b. Check Task State (conditional)

If the project uses a task tracker, check for ready, in-progress, blocked, and epic status. If no task tracker is configured, skip this step — the standup will focus on git activity and learning health only.

1c. Check Git

git log --oneline -10
git status

Note the most recent commit SHA and date. This anchors the freshness of activity signals in Phase 2.

---

Phase 2: Per-Member Activity

For each team member, gather their status without dispatching agents — standup must be fast and read-only.

2a. Git Activity by Ownership

For each member, check recent commits touching their owned files:

git log --oneline -5 -- <owns patterns>

For example, if backend owns src/domain/** and src/infra/**:

git log --oneline -5 -- "src/domain/" "src/infra/"

Record: number of commits, date of most recent commit, and whether any commits appear in the last 7 days.

2b. Learning Health

Read each member's .claude/tackline/memory/agents/<name>/learnings.md and note:

• Total entries: Number of non-empty bullet points
• Recent additions: Entries with dates in the last 7 days
• Staleness: Time since last entry was added
• Cross-agent notes: Pending notes from other members
• Size: Whether approaching the 60-line cap (30 Core + 30 Task-Relevant)

If the file does not exist, record status as cold (role not yet dispatched).

2c. Relevant Tasks (conditional)

If a task tracker is available, for each member check whether any in-progress or ready tasks match their ownership patterns by scanning task titles against the member's role keywords. If no task tracker is configured, skip this step.

---

Phase 3: Verification Gate

Before synthesizing, verify the data you gathered is complete and consistent.

Coverage check:

• Every member from team.yaml has a row of data (git + learnings). If any member is missing, go back to Phase 2 and complete their data.
• Git activity was checked against each member's actual owns patterns, not just the overall log. A member with recent commits to unowned files should still show "no owned-file activity."

Consistency check:

• If a member has recent commits but stale learnings (no entries in the last 14 days), flag this as an inconsistency: active work without captured learnings suggests a retro is overdue.
• If a member has zero owned-file commits AND zero learnings entries, mark them cold rather than stale — cold means the role hasn't been exercised yet, not that it has gone quiet.

Task tracker completeness check (if available):

• Confirm task state queries (ready, in-progress, blocked) all returned results (even if the result is zero items). If a query failed or was skipped, note "task data incomplete" in the board output.

If any check fails, complete the missing data before proceeding. Do not synthesize on partial information.

---

Phase 4: Synthesize

4a. Present the Board

## Standup: [team name]

### Team Activity
| Member | Recent Commits | Learnings | Health |
|--------|---------------|-----------|--------|
| [name] | [count] in owned files ([date of most recent]) | [total] entries ([recent] new in 7 days) | [status] |
| ... | ... | ... | ... |

Health labels:
- healthy: recent owned-file commits AND learnings active in last 7 days
- active: recent owned-file commits but learnings not updated recently (flag for retro)
- learning: new learnings entries but no recent owned-file commits
- stale: no owned-file commits AND no new learnings in last 7 days
- bloated: learnings approaching or over 60-line cap (run /curate or /retro)
- cold: no learnings file and no owned-file commits (role not yet dispatched)

### Backlog Snapshot
[Only include this section if a task tracker is available]
- **Ready**: [count] tasks available
- **In Progress**: [count] tasks active
- **Blocked**: [count] tasks blocked

### Epic Progress
[Only include if task tracker is available and epics exist]
| Epic | Children | Complete | Progress |
|------|----------|----------|----------|
| [title] | [total] | [done] | [%] |

### Blockers
[Only if task tracker is available. List any blocked tasks or issues. "No blockers" if clear.]

### Learning Highlights
[Notable recent learnings across the team. 2-3 most relevant items. "No recent learnings" if none in last 7 days.]

### Cross-Agent Notes Pending
[Any cross-agent notes that haven't been acknowledged. "None pending" if clear.]

### Flags
[Inconsistencies from Phase 3 verification: members with active commits but stale learnings, tasks with no assigned member, etc. Omit section if no flags.]

### Suggested Actions
1. [Highest-priority action based on standup data]
2. [Second priority, if any]
3. [Third priority, if any]

4b. Standup Is Complete When

The standup is complete when all of the following are true:

• Every member in team.yaml has a row in the Team Activity table
• Every health label is supported by at least one data point (commit count or learnings entry count)
• Backlog snapshot is present if a task tracker is available, or explicitly noted as unavailable
• Flags section is present if any inconsistency was detected in Phase 3, or omitted if none

If any condition is not met, complete the missing work before presenting the board.

4c. Offer Next Steps

After presenting the board, offer:

• Dispatch work on the highest-priority ready task → /sprint
• Resolve a blocker → describe the blocked task and options
• Discuss a tension or decision → /meeting
• Prune stale or bloated learnings → /retro or /curate <member>

---

Guidelines

1. Fast and read-only. Standup completes in under 30 seconds. No agent dispatch, no file writes, no task creation.
2. Honest over polished. "No activity" and "No learnings yet" are valid states. Do not interpret absence of data as presence of progress.
3. Verify before synthesizing. Phase 3 exists to catch incomplete or contradictory data before it becomes a misleading board.
4. Learning health alongside work status. A member with stale learnings and active commits is working without capturing context — that is a signal, not just a gap.
5. 60-line cap for learnings. The cap is 60 lines (30 Core + 30 Task-Relevant). Flag members at or above 50 lines as approaching the cap. Do not use 150 lines as the threshold — that is not the convention for this project.

See also: /status (solo equivalent — use when no team is configured or you want a faster individual view), /sprint (dispatch work after standup reveals what's ready), /retro (address stale learnings flagged during standup), /curate (optimize a member's learnings before the next sprint).