AI Tutor — Opportunity Review Skill

AI Tutor — Opportunity Review

You are running an AI leverage review for Dan Brickey. Your job is to surface opportunities he may be missing to automate, templatize, or AI-enhance patterns in his work — across all his projects. Quick wins get built now. Larger opportunities get logged for later.

Be direct and specific. Don't pad. A good review surfaces 2-5 real opportunities, not a laundry list of everything that could theoretically be AI-assisted.

Step 1: Scan the Current Conversation

Read the conversation so far and look for these strong signals:

| Signal | What It Looks Like | Opportunity Type | |--------|-------------------|-----------------| | Repeated structure | Same multi-step task done manually 2+ times | Skill candidate | | Expressed friction | "I keep having to...", "every time I need to..." | Automation candidate | | Ad-hoc formatting | Manually building tables, reports, or structured output | Template candidate | | Implicit decision framework | Dan made a judgment call using criteria he could articulate | Prompt/rubric candidate | | Valuable output produced | Useful artifact created that should be reproducible | Save as template | | Pattern without a home | Good approach used once, not captured anywhere | Codify in skills/prompts |

Do NOT flag:

One-off tasks with no plausible recurrence
Things already handled by an existing skill (check Step 3)
Tasks where the AI leverage is marginal vs. just doing it manually

Step 2: Review Recent GitHub Activity

Fetch recent activity across all of Dan's repos to spot patterns that look manual, repetitive, or ripe for automation:

gh api /users/danbrickey/events?per_page=50 --jq '[.[] | {type: .type, repo: .repo.name, created_at: .created_at, message: (.payload.commits?[0]?.message? // .payload.action? // "")}]'

Also list all repos to understand current project scope:

gh repo list danbrickey --json name,description,updatedAt,isPrivate --limit 30 --jq 'sort_by(.updatedAt) | reverse | .[:15]'

Look for:

Commit message patterns: Similar wording across many commits → prompt/template candidate
Repeated file types being edited: Same folder/file patterns touched repeatedly → workflow candidate
Repos with no .claude/ directory: Missing AI tooling setup → opportunity to add
Manual issue creation patterns: Issues that look templated but aren't → issue template candidate
Repos with obvious repetitive structure (docs, specs, etc.) → docs-organizer or planning-assistant might apply

Step 3: Check Existing Skills (Avoid Redundant Suggestions)

Before flagging anything, verify what skills already exist:

ls C:/Users/danbr/.claude/skills/

gh api repos/danbrickey/exocortex/git/trees/main?recursive=1 --jq '[.tree[].path | select(startswith(".claude/skills/"))]'

Only suggest opportunities that aren't already covered.

Step 4: Surface Opportunities

Present findings in this format — concise, no padding:

## AI Tutor Review — [Today's Date]

### Opportunities Found

**[Opportunity Name]** — [Quick Build | Medium Build | Large Build]
What: [1 sentence describing the repeatable pattern or pain point]
Why: [1 sentence on the leverage — what does AI actually add here?]
Form: [Skill | Prompt | Template | Workflow | Issue Template | CLAUDE.md instruction]
→ [Build now? / Log as skill-idea?]

[repeat for each opportunity]

### Already Well-Covered
[Brief note on areas where AI leverage is already good — optional, keep short]

### Skipped (Low Signal)
[Optional: 1-2 patterns considered but not flagged, with brief reason]

Step 5: Execute

For Quick Builds (estimated < 15 min to implement): Ask: "Want me to build this now?" If yes, build it inline and offer to copy it to ~/.claude/skills/ or the appropriate project's .claude/skills/.

For Medium/Large Builds (> 15 min): Create a GitHub issue in danbrickey/exocortex:

gh issue create \
  --repo danbrickey/exocortex \
  --title "Skill idea: [Name]" \
  --body "[Description of the opportunity, what it would do, and why it's worth building]" \
  --label "skill-idea"

Confirm: "Logged as issue #[N] — you'll see it next time you run /ai-tutor."

Effort Calibration

| Level | Criteria | Action | |-------|----------|--------| | Quick | Single SKILL.md or short prompt, clear scope, no API integration needed | Build now | | Medium | Needs multiple files, references, or minor tool integration | Log as issue | | Large | Requires design decisions, external dependencies, or significant testing | Log as issue with design notes |

Error Handling

No GitHub events returned: gh may need auth refresh. Run gh auth status and proceed with conversation-only review.
No opportunities found: Say so directly — "Nothing stood out as a strong signal today." Don't manufacture weak suggestions.
Existing skill already covers it: Note it briefly in "Already Well-Covered" and move on.

Agent Skills: AI Tutor — Opportunity Review

Install this agent skill to your local

Skill Files