easy-code-reader: Command for GitHub Copilot

Easy Code Reader Usage Rules

Use these two tools for Java source in a local Maven repository:

1. search_group_id: find candidate Maven groupIds and versions from a known artifact ID.
2. read_jar_source: read a fully qualified class from an explicit Maven coordinate.

Search

Supply artifact_id without the version. If known, provide a complete group_prefix, such as org.springframework; it matches group segments from the beginning, case-insensitively. version_hint is an optional version substring. Omit uncertain filters. Every search reads the current repository directly, so newly installed or removed dependencies appear on the next call.

Choose the intended groupId and version from matches. Numeric version segments follow Maven ordering; do not infer the desired application version solely from the first result.

Read

{
  "group_id": "org.springframework",
  "artifact_id": "spring-core",
  "version": "5.3.21",
  "class_name": "org.springframework.core.SpringVersion"
}

Omitted line bounds return the complete source. To reduce subsequent output, pass start_line and/or end_line; lines are 1-based and both bounds are inclusive. For an explicit range, inspect total_lines and is_partial before treating a snippet as the whole implementation. Those fields are omitted for full reads.

Internal classes use binary names, such as org.example.Outer$Inner. The tool may return the outer file with a short warning. Full reads otherwise return only class_name, artifact, source_type and code.

Sources are preferred unless prefer_sources=false. SNAPSHOTs prefer the ordinary main JAR for reading and the latest timestamped main JAR name for caching, with numeric build-number ordering. If the ordinary main JAR is absent, read the timestamped JAR. Ordinary main JARs prefer ordinary sources, then matching timestamped sources; timestamped main JARs use matching sources. The actual path and resolved version are kept in internal diagnostics rather than repeated in every response.