v0.7

What's Changed

• Security container hardening by @AdminTurnedDevOps in #1
• Fix retry adjustments, rename max_retries->max_attempts, wire cleanup by @AdminTurnedDevOps in #2
• Parse pytest JUnit XML instead of scraping stdout by @AdminTurnedDevOps in #3
• Align docs with implementation; untrack compiled binary by @AdminTurnedDevOps in #4
• Add run telemetry: whole-pipeline tokens, duration, and cost by @AdminTurnedDevOps in #5
• Deduplicate CodeScorer scoring methods by @AdminTurnedDevOps in #6
• Remove default models; require an explicit model by @AdminTurnedDevOps in #7

Full Changelog: v0.6...v0.7

v0.6

Fixes to improve performance for task completion.

v0.05

MCP registry search full implementation

v0.4

Implemented Agent Skill search functionality so validtr goes and searches Agent Skill catalogs/registries and tells you the best Skill to use for your stack.

v0.3

Got the output that I wanted to see for a prototype.

v0.2

output mod

v0.1

output