feat: v0.7.1

[kiloloop-release]

What's New in v0.7.1

Fixed

• Fixed six engine correctness gaps: early --format validation before side effects, explicit --warm-context 1.0 handling, friction-aware assigned-agent METR re-checks, Python 3.10-compatible Z timestamp parsing, UTC-normalized calibration cohorts/order, and plan_waves duplicate/negative input validation (Add Claude Code plugin structure #31, Add Codex .agent skill mirror #33, Fix METR threshold lookup: claude agent should resolve to opus (p80=90m) #36, Auto-detect warm context from dispatch history file (data.json integration) #37, Issue-mode: allow modifier flags to apply uniformly across all fetched issues #38, Calibrate modifier floors and fix METR key resolution #39).
• README hero example now uses the real CLI surface with real captured output — the previous block invoked a nonexistent --model flag, passed an unsupported second positional task, and showed fabricated numbers (Add integration tests, regression fixtures, and CI workflow #28).
• README Action workflow example now grants issues: read, required for issue fetching on private repos where an explicit permissions: block zeroes unlisted scopes (Add estimate CLI flags for modifier control #40).
• Claude Code plugin exposes the /estimate skill again: plugin.json now points skill discovery at skills/estimate/claude/, which the v0.7.0 multi-runtime restructure had moved out of the default skills/<name>/SKILL.md scan path (Add comprehensive unit tests for core modules #29).
• Dropped the phantom /validate-estimate and /calibrate slash commands (documented but never installable); validation and calibration now route through the installed skill as /estimate validate and /estimate calibrate (Add comprehensive unit tests for core modules #29).
• Refreshed stale METR model keys in examples/multi-agent.md output (gpt_5_3 → gpt_5_4, gemini_3_pro → gemini_3_1_pro) to match current output (Add integration tests, regression fixtures, and CI workflow #28).