archetech / archon / 30681420334

93%

main: 93%

Pull #816

github

test: resolve @didcid/cipher/didcomm to source, not the built dist

jest.config.js maps @didcid/cipher/node, /types and /passphrase to
TypeScript source but had no entry for /didcomm, so the import in
packages/keymaster/src/keymaster.ts resolved through package exports to
packages/cipher/dist/esm/didcomm.js.

The DIDComm logic was therefore measured twice — dist at 83.6% and src at
96.4% — inflating the denominator with build output, and coverage
annotations pointed at generated JS rather than actionable source. Local
runs also exercised whatever was last compiled rather than the working
tree (CI builds first, so it was compiling fresh).

Verified before switching: src and dist export identical symbols, no
commit has touched packages/cipher/src/didcomm.ts since the dist was
built, and dist/ is gitignored.

Build output is now absent from coverage entirely (7607 -> 7394 measured
lines) and src/didcomm.ts rises to 97.4% as the tests that were hitting
dist now exercise source. Repo total 96.19% -> 96.58%.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

2998 of 3510 branches covered (85.41%)

Branch coverage included in aggregate %.

7140 of 7394 relevant lines covered (96.56%)

768.32 hits per line