Skip to content
GitWhy

Live demo report

This page shows what gitwhy index && gitwhy status && gitwhy risk looks like when run against the gitwhy repository’s own history. It’s the same data your editor would see through the MCP tools.

The capstone demo video walks through this report live; the static snapshot here serves both as a fallback for readers without an MCP-capable editor and as proof that the pipeline produces sensible output on real code.

What was indexed

Terminal window
$ gitwhy init
✓ initialized gitwhy at .gitwhy
  commits in repo: 5


$ gitwhy estimate
Estimate for 5 commits (model: gemini-2.5-flash)
Category       Count   LLM calls   Tokens (prompt/completion)   Est. cost
-------------------------------------------------------------------------
mega           4       16          92800/640                    $0.0294
initial        1       0           0/0                          $0.0000
-------------------------------------------------------------------------
TOTAL          5       16          92800/640                    $0.0294


$ gitwhy index
indexed 5/5 (enriched=4, skipped=0, errors=0, est. $0.0277)
done in 41s — processed 5, enriched 4, errors 0, cost $0.0277

Index status

Terminal window
$ gitwhy status
Indexed: 5/5 commits  (100% coverage)
Embeddings: 4
LLM usage: 34 calls, 91038+180 tokens, $0.0277
Last indexed: 2026-05-15T08:43:04.340Z
DB size: 156 KB


Top hotspots (last 30 days):
  src/cli/index.ts  (5 commits)
  src/index.ts  (4 commits)
  package.json  (3 commits)
  pnpm-lock.yaml  (3 commits)
  src/mcp/runtime.ts  (3 commits)

Risk assessment for a single file

Terminal window
$ gitwhy risk src/indexer/indexer.ts


MEDIUM risk (score 0.42)  —  src/indexer/indexer.ts


Reasons:
  • bus factor = 1 (kambar owns 100% of the file)
  • only one contributor in indexed history


Stats: bus factor 1, 1 contributors, 2 commits, 2 in last 90 days.


Top contributors:
  kambar                    100%  2 commits, last 2026-05-15

This is the canonical capstone-demo example: GitWhy correctly identifies that src/indexer/indexer.ts has a single contributor (solo project) and flags this as MEDIUM-risk because bus-factor-of-1 means a single point of failure for the file’s institutional knowledge.

Co-changing files

Terminal window
$ gitwhy related src/indexer/indexer.ts --min 1


Files that change with "src/indexer/indexer.ts":
  src/cli/index.ts                                   2/2 commits  (100%)
  tests/unit/storage.test.ts                         1/2 commits  (50%)
  tests/unit/sqlite-blob-vector-store.test.ts        1/2 commits  (50%)
  tests/unit/secret-detection.test.ts                1/2 commits  (50%)
  tests/unit/ping.test.ts                            1/2 commits  (50%)

The 100% forward-confidence on src/cli/index.ts is meaningful: every commit that touched the indexer also touched the CLI entry. If you change the indexer’s behavior, expect to update the CLI too.

MCP-doctor diagnostic

Terminal window
$ gitwhy mcp-doctor --no-probe
✓ .gitwhy/ initialized
✓ Config valid (provider=gemini, indexingModel=gemini-2.5-flash, ...)
✓ Index populated (5 commits, 4 embeddings)
✓ MCP tools registered (9 tools: gitwhy.ping, gitwhy.why, gitwhy.history,
  gitwhy.search, gitwhy.catchup, gitwhy.risk, gitwhy.related,
  gitwhy.context_for_pr, gitwhy.suggest_commit_message)
✓ LLM credentials (provider=mock)


Summary: 5 ok, 0 warn, 0 fail.

What’s not shown here

  • The MCP-native split-screen demo (the headline moment) — that’s in the video; see the README for the link once recorded.
  • A larger demo repo — running gitwhy against tj/commander.js or a similar mid-size OSS repo produces a much richer report. That bundle is being prepared as part of the OSS launch.
  • Real LLM cost numbers — the report above used Gemini’s free tier; numbers will differ for paid OpenAI / Gemini tiers.