coverage-report

name	coverage-report
description	Run transformation coverage tests. Use compact mode when iterating on fixes, full mode when planning or documenting bugs.

Coverage Report Skill

Run cross-provider transformation coverage tests for Lingua.

When to Use Each Mode

Compact Mode (-f compact) - Use When Iterating

Use compact mode to save tokens (~95% smaller output) when:

• Fixing a specific bug and checking if it's resolved
• Running quick validation after code changes
• Iterating rapidly on implementations
• You already understand the failure patterns

# Quick check after a code change
cargo run --bin coverage-report -- -f compact

# Filter to specific providers
cargo run --bin coverage-report -- -f compact -p anthropic,responses

# Filter to specific test case
cargo run --bin coverage-report -- -f compact -t seedParam

Full Markdown Mode - Use When Planning/Documenting

Use full mode (default) when:

• Initial investigation of failures (need full error details and diffs)
• Writing bug reports to .claude/BUGS.md or .claude/*_bugs.md
• Planning implementation approach
• Need to understand the scope of issues

# Full report for planning
cargo run --bin coverage-report

# Save to bug tracking file
cargo run --bin coverage-report -- -p anthropic > .claude/anthropic_bugs.md

Filtering Options

Option	Short	Description
--coverage		requests, responses, streaming, roundtrip, all
--test-cases	-t	Test case patterns (glob: * any chars, ? single char)
--providers	-p	Provider filter (both source AND target must match)
--source		Filter source providers only
--target		Filter target providers only
--format	-f	markdown (default), compact

Common Filter Patterns

# Only request transformations
cargo run --bin coverage-report -- -f compact --coverage requests

# Test cases matching pattern
cargo run --bin coverage-report -- -f compact -t "reasoning*"
cargo run --bin coverage-report -- -f compact -t "*Param"

# Specific direction
cargo run --bin coverage-report -- -f compact --source anthropic --target responses

Provider Abbreviations (Compact Mode)

In compact output, providers are abbreviated:

• oai = ChatCompletions (OpenAI)
• ant = Anthropic
• ggl = Google
• bed = Bedrock
• rsp = Responses (OpenAI Responses API)

Reading Compact Output

# Coverage (compact)
Stats: 669/1704 (39.3%) [512+157lim] 1035fail
req:617/836 res:32/424 str:20/444

## Failures (79 patterns, 1035 total)

[P1] L:usage.prompt_cache_creation_tokens (123)
  ant→ggl: cacheControl1hParam (response)...(+44)

• L: = Lost fields (present in source, missing after transform)
• A: = Added fields (not in source, appeared after transform)
• C: = Changed fields (value differs after transform)
• (123) = Number of test cases with this pattern
• ...(+44) = 44 more test cases not shown

Workflow Example

1. Initial investigation (full mode):

   cargo run --bin coverage-report -- -p anthropic,responses > .claude/ant_rsp_bugs.md
   

2. Implement fix and iterate (compact mode):

   cargo run --bin coverage-report -- -f compact -p anthropic,responses
   

3. Verify fix and update documentation (full mode):

   cargo run --bin coverage-report -- -p anthropic,responses
   # Update .claude/BUGS.md with results