2 commit fast

Fast Commit

Analyze staged changes and commit instantly with a well-crafted conventional commit message. No confirmation step — speed is the point.

Step 1: Check for staged changes

Run git diff --cached --stat to see what's staged.

If nothing is staged, check git status for unstaged changes. If there are unstaged changes, tell the user nothing is staged and suggest they stage files first. Do not run git add unless the user explicitly asks you to. If the repo is clean, say so and stop.

Step 2: Gather context

Run these in parallel:

• git diff --cached — the full staged diff (this is the primary input for the commit message)
• git log --oneline -20 — recent commit history, so you can match the project's existing scope conventions
• git branch --show-current — the branch name often encodes intent (e.g., fix/null-user-lookup, feat/oauth-google)

Conversation context: Before analyzing the diff, review what the user has been working on in this conversation. Tasks they described, bugs they debugged, or features they built are the strongest signal for the why behind a change. Prefer conversation context over guessing intent from code alone.

Step 3: Analyze the diff

Determine what changed and why:

1. Identify the what: Which functions, classes, routes, or config keys were added, modified, or deleted? Which single module, package, or component is affected (this becomes the scope)?
2. Determine the why using these signals (in priority order):
   • Conversation context — what was the user trying to accomplish? This overrides all heuristics below.
   • Branch name — fix/, feat/, chore/ prefixes directly indicate type and intent.
   • Diff patterns that reveal intent:
     • Conditional/guard clause added or changed around existing logic → likely fix
     • Return value or error handling changed → likely fix
     • New exported function/class/route/component → likely feat
     • Renamed symbols, moved code between files, extracted helpers with no new behavior → refactor
     • Only .test. / .spec. / __tests__ files changed → test
     • Only package.json, lockfile, .config files → chore or build
3. Gauge complexity: Count files changed and total lines added/removed.

Large diffs (>300 lines or >8 files)

For big diffs, don't try to describe everything. Instead:

• Identify the single unifying intent (e.g., "migrate from REST to GraphQL", "add user profile feature")
• If there is no unifying intent, describe the dominant change and note the secondary ones in the body
• Skim file names and hunks for the theme; don't get lost in implementation details

Step 4: Build the commit message

Format

Follow Conventional Commits:

type(scope): description

[optional body]

For breaking changes, append ! after the type/scope:

type(scope)!: description

What breaks and how to migrate.

A breaking change always requires a body explaining what breaks.

Type

Pick the type that best describes what the change does:

Type	Use when
feat	New user-facing functionality
fix	Bug fix
docs	Documentation only (README, comments, docstrings)
style	Formatting, whitespace — no logic change
refactor	Code restructuring with no behavior change
test	Adding or updating tests
chore	Maintenance: deps, configs, build scripts, tooling
perf	Performance improvement
ci	CI/CD pipeline changes
build	Build system or external dependency changes
revert	Reverting a previous commit

Scope

Look at the recent commit log from Step 2 to understand what scopes the project already uses. Match those conventions — if the project uses auth, don't invent authentication. If the project has never used scopes, don't start now.

Rules:

• Scope should identify a single area of the codebase (a module, package, directory, or component)
• If the change touches multiple areas that don't share a single natural scope, omit the scope entirely — write type: description instead of type(scope1, scope2): description
• Never use comma-separated or slash-separated scopes

Subject line

• Lowercase, imperative mood ("add", not "added" or "adds")
• 72 characters max for the entire subject line (including type(scope): )
• Describe what the change does, not which files were touched
• Be specific: fix null pointer in user lookup beats fix bug

Anti-patterns — never generate these:

• Vague: "update code", "fix issue", "improve handling", "make changes", "address feedback"
• Tautological: "refactor: refactor auth module", "fix: fix the bug"
• File-listing: "update user.ts and auth.ts"
• Over-broad: "improve application" or "update project"

If you catch yourself writing a vague subject, ask: what specifically was broken/missing/wrong? Name the concrete thing.

Body

Add a body (1-3 sentences) when any of these are true:

• The change is a fix and the subject doesn't name the root cause
• The change is a breaking change (always explain what breaks)
• The diff touches >5 files or >150 lines
• The why is not obvious from the what (e.g., a performance fix that changes algorithm)

Skip the body when the subject fully explains a small, obvious change (rename, dep bump, typo fix).

Step 5: Commit

Run the commit immediately:

git commit -m "type(scope): description"

Or with a body:

git commit -m "$(cat <<'EOF'
type(scope): description

Brief explanation of why, if needed.
EOF
)"

After committing, show the user the resulting commit hash and message.

Examples

Single-scope changes:

feat(auth): add OAuth2 integration with Google provider
fix(api): resolve timeout on token refresh
docs(readme): update installation instructions for v2.0
chore(deps): update express to 4.19.2
test(auth): add tests for token expiration edge cases

Multi-scope changes (scope omitted):

refactor: rename internal helpers for consistency
chore: update linter config and CI pipeline
feat: add user profile page with API endpoint

Breaking change:

feat(api)!: require API key for all endpoints

With body:

fix(parser): handle unterminated string literals

Previously, an unterminated string caused an infinite loop in the tokenizer.
The parser now emits a diagnostic and recovers at the next newline.

Important

• Do not ask the user to confirm the message — just commit
• Do not generate multiple options — pick the best one
• Do not include co-authorship footers or AI attribution
• Do not run git add unless the user explicitly asked you to stage files
• Do not commit if nothing is staged