Brag Sheet

Brag sheet

Turn scattered engineering activity into concise, evidence-backed impact entries by mining GitHub Copilot CLI sessions, git commits, and GitHub pull requests, then enforcing action → result → evidence.

When to invoke

• "What did I do last week?"
• "Backfill my work history for this half."
• "Write impact statements for my performance review."
• "Summarize what I shipped from my PRs."
• "Help me prep a promo packet or weekly update."

Inputs

Use $ARGUMENTS for the time range, category, mode, or scope. If it is ambiguous, confirm the range and source scope before scanning. Common modes are Capture, Backfill, and Review Pack.

Limits

• Do not use this skill for project management, sprint planning, time tracking, or ticket creation.
• Do not fabricate metrics, team size, impact numbers, or shipped status.
• Do not save entries with save_to_brag_sheet unless that tool is confirmed available and the user approves the drafted entries.

Modes

Impact contract

Every entry must include all three parts:

Did [action] → [result/impact] → [evidence]

If evidence is missing, write (evidence needed) and ask for proof. Never silently omit evidence.

Evidence ladder

Qualitative evidence with context beats fabricated numbers.

Categories

Procedure

1. Confirm time range and scope. Do not assume "last week" without dates.
2. Check available tools and sources: save_to_brag_sheet, git, gh, and Copilot session logs.
3. For Backfill, scan all available sources before drafting entries.
4. Group related signals: same PR plus commits, multiple commits on same feature within 3 days, and Copilot sessions on the same repo and branch.
5. Draft entries using action → result → evidence and assign categories.
6. Show drafted entries before saving or finalizing.
7. For Review Pack, select the top 3-5 highest-impact items, organize by impact theme, and expand selected items using STAR.

Backfill commands

git --version 2>/dev/null
gh --version 2>/dev/null
ls ~/.copilot/session-state/ 2>/dev/null

git log --author="$(git config user.email)" --since="2 weeks ago" \
  --pretty=format:'%h|%ad|%s' --date=short --no-merges

gh pr list --author @me --state merged --limit 20 \
  --json number,title,repository,mergedAt

For Copilot session history, read ~/.copilot/session-state/<session-id>/workspace.yaml and use summary, cwd, repository, and branch. Skip sessions without summary. If ~/.copilot/session-state/ does not exist, continue with git and PRs.

For long review ranges, prefer PR history and explicit dates:

git log --author="$(git config user.name)" --since="2024-07-01" --until="2025-01-01" --oneline

Review prep

Use STAR for longer narratives: Situation → Task → Action → Result. For Microsoft Connect-style reviews, frame around Core Priorities: delivering results, customer obsession, teamwork, and growth mindset.

Gotchas

• No recent commits is not no impact: check gh pr list --author @me --state merged, other repos, other branches, or fall back to guided interview.
• Review periods may span 6-12 months: set explicit --since and --until dates.
• Pair programming needs attribution: ask whether to credit as individual work, shared work, or skip.
• Ambiguous "brag" may be launch copy: confirm whether the user wants a work entry or a team announcement.
• Weak periods should stay honest: do not pad with trivial entries.

Related primitives

Work-history terminology

Keep these terms available when mining or grouping sources: auto-save, co-authored, cross-repo, file/feature, gh pr list --state merged, git/PRs/sessions, incidents/quarter, and self-review.

Output template

## Brag sheet result

**Status:** captured | backfilled | review-pack | needs evidence | blocked
**Time range:** <explicit range>
**Sources scanned:** git | GitHub PRs | Copilot sessions | interview

### Week of <YYYY-MM-DD>

#### PRs & Features
- **<action>** → <result/impact> → <evidence>

#### Infrastructure
- **<action>** → <result/impact> → <evidence or `(evidence needed)`>

### Review themes
| Theme | Strongest entries | Evidence gaps |
| --- | --- | --- |
| `<theme>` | `<entries>` | `<missing proof>` |

### Validation
- Entries follow action → result → evidence: pass | fail
- Metrics source-verified or user-provided: pass | fail
- Draft shown before saving: yes | no

Quality gate

• Time range and scope are explicit.
• Every entry has action → result → evidence or (evidence needed).
• No metric, shipped status, or beneficiary is invented.
• Related commits, PRs, and sessions are grouped instead of duplicated.
• Draft entries are shown before saving.
• Missing tools or missing ~/.copilot/session-state/ are handled without error.
• Categories are assigned for pasteable review or weekly update output.

References

• copilot-brag-sheet