coderaiser / putout-editor / 36633604969

100%

push

github

docs: the two new mcp tools, the new rule, and an unfiled idea

Four things were stale after the last few commits, all found by checking
rather than assuming.

AGENTS.md and packages/mcp/README.md list the mcp tools, and both were
missing name_pattern - so an agent reading the table would not have known the
inverse of test_pattern exists. The README workflow now runs it before
test_pattern, since the usual starting point is code you have and a pattern you
do not.

architecture.md described one code rule and two filesystem ones. It now names
apply-linked-pattern-value and remove-comments as code rules too, and says
that a rule here can be type-checked, because putout re-exports the plugin
contract and the files stay .js only to match the 116 upstream plugins.

ideas.md gains the assertion finding, which was measured and never written
down: 201 t.ok calls, dominated by t.ok('field' in schema.shape) and
t.ok(text.startsWith('Error:')). Both are booleans that report whether and
never what, so a schema test with three field checks gives the same message
whichever one is missing. It belongs to @putout/plugin-tape, which already
ships convert-ok-to-match and convert-ok-to-pass; what is missing is the
assertion that cannot be converted, because t.ok(a && b) needs a human to
choose the expected value.

Also recorded there: tape/convert-equal-to-ok pushes the other way, and the
two rules are only in tension while a test asserts a boolean. Comparing the
value resolves it, because then neither has anything to say.

Diagrams up to date, putout clean, and every doc link and anchor checked -
including the one my slugger flagged, resolved against the real
github-slugger.

1081 of 1081 branches covered (100.0%)

Branch coverage included in aggregate %.

6670 of 6670 relevant lines covered (100.0%)

61.48 hits per line