comprehensive-analysis

# COG Comprehensive Analysis Skill

## When to Invoke
- User wants a weekly review or retro prep
- User says "weekly analysis", "comprehensive review", "board prep", "deep dive"
- User needs analysis across a longer time period (7+ days)
- User wants a full picture of team/product health

## Agent Mode Awareness

**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
- If `agent_mode: team` — use the full parallel agent execution strategy below (5 agents). This skill benefits greatly from team mode.
- If `agent_mode: solo` — run a lighter version: collect GitHub + Linear data sequentially, skip PostHog deep analysis, produce a single combined report instead of 3 documents.

## Purpose
Generate a deep-dive analysis across all data sources for weekly reviews, board prep, strategic planning, or any time you need the full picture. Unlike the team brief (which optimizes for speed and daily relevance), this skill intentionally pulls more data and spends more time synthesizing.

**Use this when:**
- Weekly team review / retro prep
- Board meeting or leadership update prep
- Strategic planning sessions
- Investigating a specific problem across all data sources
- Monthly/quarterly health check

**Don't use this daily** — it takes 8-12 minutes and pulls heavy payloads. Use `/daily-brief` for everyday intel.

## Command: `/comprehensive-analysis`

## Voice & Tone

Same as the daily brief — direct, opinionated, teammate energy. But with more depth and nuance. You're writing for a product leader who needs to make decisions, not just stay informed.

---

## Execution Strategy

### Phase 1: Deep Data Collection (~3-5 min)

**Launch ALL agents in parallel using the Task tool with `run_in_background: true`.**

#### Agent 1: "github-deep-analyst" (subagent_type: general-purpose)
```
Deep GitHub analysis for [CUSTOMIZE: your-org/your-repo].
Analysis period: last 7 days (from [7_DAYS_AGO] to [TODAY]).

Collect:
1. ALL PRs merged in the last 7 days:
   gh pr list --repo [CUSTOMIZE: your-org/your-repo] --state merged --search "merged:>=[7_DAYS_AGO]" --json number,title,author,mergedAt,labels,additions,deletions --limit 100

2. ALL open PRs with full detail:
   gh pr list --repo [CUSTOMIZE: your-org/your-repo] --state open --json number,title,author,createdAt,reviewDecision,labels,updatedAt,additions,deletions --limit 100

3. Contributor activity breakdown:
   For each contributor, count: PRs merged, PRs opened, commits, review comments given.
   gh api repos/[CUSTOMIZE: your-org/your-repo]/stats/contributors

4. Code frequency (additions/deletions per week):
   gh api repos/[CUSTOMIZE: your-org/your-repo]/stats/code_frequency

5. Commit activity by day:
   gh api repos/[CUSTOMIZE: your-org/your-repo]/stats/commit_activity

6. ALL PR review comments from the last 7 days:
   gh api repos/[CUSTOMIZE: your-org/your-repo]/pulls/comments --paginate --jq '[.[] | select(.created_at >= "[7_DAYS_AGO]")]'

DEEP ANALYSIS:
A) **Contributor Velocity Matrix**: For each contributor — PRs merged, lines changed, review comments given/received. Who's shipping? Who's reviewing? Who's doing both?
B) **Code Churn**: Are we rewriting the same files repeatedly? Flag files with >3 PRs touching them in one week.
C) **PR Lifecycle**: Average time from PR open → first review → merge. Where's the bottleneck?
D) **Review Quality**: Are reviews substantive (comments with suggestions) or rubber stamps (approved with no comments)?
E) **Week-over-Week Trends**: Compare this week's velocity to last week. Accelerating or decelerating?
F) **Technical Debt Signals**: Large PRs with no tests, PRs that touch >10 files, dependency-only PRs.

Return structured data AND insights.
```

#### Agent 2: "slack-deep-monitor" (subagent_type: general-purpose)
*Only spawn if Slack MCP is available*
```
Deep Slack analysis for last 7 days across key channels.
Analysis period: [7_DAYS_AGO] to [TODAY].

Instructions:
1. Use ToolSearch to load Slack tools
2. Read messages from [CUSTOMIZE: your-team-channel] for the full 7-day window
3. If you have access, also check: [CUSTOMIZE: additional-channels] (e.g., #general, #engineering, #product)

For EACH significant thread (>3 replies or involving decisions):
- Full topic summary
- Key participants
- Decision reached (or explicitly: "no decision reached")
- Action items with owners
- Sentiment (positive/negative/neutral/heated)
- Links to external resources shared

DEEP ANALYSIS:
A) **Communication Patterns**: Who's driving discussions? Who's mostly silent? Any asymmetry?
B) **Decision Velocity**: How many decisions were made vs. how many discussions ended without resolution?
C) **Topic Clustering**: Group discussions by theme (product, engineering, bugs, strategy, etc.)
D) **Unresolved Threads**: List all discussions that need follow-up — no decision, open question, blocked waiting for someone
E) **Sentiment Map**: Overall team mood. Are discussions constructive or frustrated?
F) **External Intel**: All links shared (articles, competitor news, tools) — categorize by relevance

Return structured data AND insights.
```

#### Agent 3: "linear-deep-tracker" (subagent_type: general-purpose)
*Only spawn if Linear MCP is available*
```
Deep Linear analysis for last 7 days.
Analysis period: [7_DAYS_AGO] to [TODAY].

Instructions:
1. Use ToolSearch to load ALL Linear tools
2. Collect comprehensive data:

   a) All issues updated in last 7 days (mcp__claude_ai_Linear__list_issues)
   b) All issues created in last 7 days
   c) All issues completed in last 7 days
   d) All blocked issues (any date)
   e) Full initiative list with projects (mcp__claude_ai_Linear__list_initiatives)
   f) Detailed status for each initiative (mcp__claude_ai_Linear__get_initiative)
   g) All active cycles (mcp__claude_ai_Linear__list_cycles)
   h) All milestones for active projects (mcp__claude_ai_Linear__list_milestones)
   i) All projects with progress (mcp__claude_ai_Linear__list_projects)
   j) Recent status updates (mcp__claude_ai_Linea