Skip to content

Latest commit

 

History

History
82 lines (60 loc) · 3.22 KB

File metadata and controls

82 lines (60 loc) · 3.22 KB

Repository Summary

  • Entities: 0
  • Relations: 0
  • Changes: 0

Kin-First Workflow

This repository uses Kin semantic VCS. Kin indexes every function, class, type, and trait into a graph with cross-file relations. Use Kin tools instead of raw file operations whenever possible.

Finding Code (use Kin instead of grep/rg/find)

Instead of Use Why
grep -r function_name kin search function_name Returns exact entity definitions with file:line, not noisy text matches
rg 'class Foo' kin search Foo --kind class Disambiguates: shows only definitions, not usages
find . -name '*.rs' then read each kin search --language rust Queries the semantic graph, no filesystem walk

Reading Code (use Kin instead of cat/read entire files)

Instead of Use Why
Reading 5 files to find callers kin context <entity> Token-budgeted pack: focal entity + callers + dependencies
cat src/foo.rs (whole file) kin search foo then read only the pointed-to file 84% fewer tokens on average
Guessing which files matter kin support Coverage report showing all indexed entities, languages, relations

Reviewing Changes (use Kin instead of git diff)

Instead of Use Why
git diff kin review Entity-level diff + downstream impact + risk assessment
Manual impact guessing kin review impact section Shows callers, dependents, contracts, tests affected

Committing (use Kin for semantic history)

Instead of Use Why
git commit kin commit -m "message" Tracks entity-level changes, not line diffs
git log kin history Semantic change history per entity

Quick Reference

kin search <pattern>          # Find entities (functions, classes, types)
kin search <p> --kind <k>     # Filter by kind: function, class, struct, trait
kin search <p> --language <l>  # Filter by language: rust, typescript, python, go, java
kin context <entity>          # Token-budgeted context pack
kin support                   # Coverage report (entities, languages, relations)
kin review                    # Semantic diff + impact + risk assessment
kin commit -m "msg"           # Semantic commit
kin history                   # Entity change history
kin work list                 # Outstanding work items
kin verify <entity>           # Check test coverage for an entity

Key Principle

Search semantically first, read files second. Kin search returns 4-22x less noise than grep and context packs use 84% fewer tokens than reading all matching files. Only fall back to raw file reads when Kin doesn't have what you need.

Conventions

  • Use kin commit -m "message" for semantic commits
  • Use kin context <entity> for token-budgeted context packs
  • Use kin review for semantic impact analysis before merging
  • Use kin search <pattern> to find entities across the codebase

Verification

  • Work items: 0
  • Use kin verify <entity> to check test coverage for an entity
  • Use kin work list to see outstanding work items