status（状态）

Status Reporting

Gather -> aggregate -> present pattern for comprehensive project status across VCS, PRs, issues, CI.

<when_to_use>

• Starting work sessions (context refresh)
• Checking project/team activity
• Understanding PR/stack relationships
• Quick status overview before planning
• Reviewing recent changes across systems
• Understanding blockers

NOT for: deep-dive into specific items, real-time monitoring, single-source queries

</when_to_use>

<core_pattern>

Three-stage workflow:

1. Gather - collect from multiple sources
2. Aggregate - combine, filter, cross-reference by time/stack/status
3. Present - format for scanning with actionable insights

Key principles:

• Multi-source integration (VCS + code review + issues + CI)
• Time-aware filtering (natural language -> query params)
• Stack-aware organization (group by branch hierarchy)
• Scannable output (visual indicators, relative times)
• Actionable insights (highlight blockers, failures)

</core_pattern>

<workflow>

Stage 1: Parse Constraints

Extract time from natural language:

• "last X hours" -> -Xh
• "past X days" / "last X days" -> -Xd
• "yesterday" -> -1d
• "this morning" / "today" -> -12h
• "this week" -> -7d
• "since {date}" -> calculate days back

Default: 7 days if unspecified.

Stage 2: Gather Data

Run parallel queries for each available source:

1. VCS State - branch/stack structure, recent commits, working dir status
2. Code Review - open PRs, CI status, review decisions, activity
3. Issues - recently updated, status, priority, assignments
4. CI/CD - pipeline runs, success/failure, error summaries

Skip unavailable sources gracefully.

Stage 3: Aggregate

Cross-reference and organize:

• Group PRs by stack position (if stack-aware)
• Filter all by time constraint
• Correlate issues with PRs/branches
• Identify blockers (failed CI, blocking reviews)
• Calculate relative timestamps

Stage 4: Present

Format for scanning:

• Hierarchical sections (VCS -> PRs -> Issues -> CI)
• Visual indicators (✓ ✗ ⏳ for status)
• Relative timestamps for recency
• Highlight attention-needed items
• Include links for deep-dive

See templates.md for section formats.

</workflow>

<data_sources>

VCS - stack visualization, commit history, working dir state

• Stack-aware (Graphite, git-stack): hierarchical branch relationships
• Standard git: branch, log, remote tracking

Code Review - PRs/MRs, CI checks, reviews, comments

• Platforms: GitHub, GitLab, Bitbucket, Gerrit

Issues - recent updates, metadata, repo relationships

• Platforms: Linear, Jira, GitHub Issues, GitLab Issues

CI/CD - runs, success/failure, timing, errors

• Platforms: GitHub Actions, GitLab CI, CircleCI, Jenkins

Tool-specific: graphite.md, github.md, linear.md, beads.md

</data_sources>

<aggregation>

Cross-Referencing:

1. PRs to branches (by name)
2. Issues to PRs (by ID in title/body)
3. CI runs to PRs (by number/SHA)
4. Issues to repos (by reference)

Stack-Aware Organization:

• Group PRs by hierarchy
• Show parent/child relationships
• Indicate current position
• Highlight blockers in stack order

Filtering:

• Time: apply to all sources, use most recent update
• Status: prioritize action-needed, open before closed

Relative Timestamps:

• < 1 hour: "X minutes ago"
• < 24 hours: "X hours ago"
• < 7 days: "X days ago"
• >= 7 days: "X weeks ago" or absolute

</aggregation>

<presentation>

Visual Indicators:

• ✓ success | ✗ failure | ⏳ pending | ⏸ draft | 🔴 blocker
• ▓▓▓░░ progress (3/5)
• ◇ minor | ◆ moderate | ◆◆ severe

Output Structure:

=== STATUS REPORT: {repo} ===
Generated: {timestamp}
{Time filter if applicable}

{VCS_SECTION}
{PR_SECTION}
{ISSUE_SECTION}
{CI_SECTION}

⚠️ ATTENTION NEEDED
{blockers and action items}

See templates.md for detailed section templates.

</presentation>

<scripts>

Use scripts/sitrep.ts for automated gathering:

./scripts/sitrep.ts              # All sources, 24h default
./scripts/sitrep.ts -t 7d        # Last 7 days
./scripts/sitrep.ts -s github    # Specific sources
./scripts/sitrep.ts --format=text

Outputs JSON (structured) or text (human-readable). Reduces agent tool calls 80%+.

See implementation.md for script structure and patterns.

</scripts>

<dependencies>

Required: VCS tool (git, gt, jj), shell access

Optional (graceful degradation):

• Code review CLI (gh, glab)
• Issue tracker MCP/API
• CI/CD platform API

Works with ANY available subset.

</dependencies>

<rules>

ALWAYS:

• Parse time constraints before queries
• Execute queries in parallel
• Handle missing sources gracefully
• Use relative timestamps
• Highlight actionable items
• Provide links for deep-dive
• Format for scanning

NEVER:

• Fail entirely if one source unavailable
• Block on slow queries (use timeouts)
• Expose credentials
• Dump raw data without organization

</rules>

<integration>

As session starter:

1. Generate report (understand state)
2. Identify attention-needed items
3. Plan work (prioritize by blockers)
4. Return periodically (track progress)

Cross-skill references:

• Failing CI -> debugging
• Before planning -> use report for context
• When blocked -> check dependencies

Automation: daily standup, pre-commit hooks, PR creation context

</integration>

<references>

Tool integrations:

• graphite.md - Graphite stack and PR queries
• github.md - GitHub CLI patterns
• linear.md - Linear MCP integration
• beads.md - Local issue tracking

Implementation:

• templates.md - Output templates and formatting
• implementation.md - Patterns, scripts, anti-patterns

Examples:

• EXAMPLES.md - Usage examples and sample output

Formatting:

</references>