v1.9 Hardening

Release v1.9

Ecosystem Alignment and Advanced Orchestration: spec-aligned skill generation, progressive disclosure, 5-platform portability, evaluator-optimizer with A/B testing, MCP distribution, router and map-reduce topologies, session save/restore, agentic RAG, and CLI improvements. Nine capabilities in a single release — the system's broadest expansion.

Structured Data

Release Metadata

Version	`v1.9`
Release Era	Hardening
New Capabilities	9
Cumulative Capabilities	30
Cumulative Tests	—

Era Context

Hardening Era

Pipeline architecture, security hardening, spec alignment, and ecosystem robustness.

Progression Metrics

Capabilities	30 of 139
Era Progress	v1.9 of v1.8–v1.10
Growth	+9 caps — largest single release

Capabilities Introduced

Architectural Decisions

Full Claude Code spec compliance — $ARGUMENTS injection, context:fork, dual-format allowed-tools, shell injection prevention; skills generated by the system are now indistinguishable from hand-authored spec-compliant skills
Progressive disclosure via SKILL.md + references/ + scripts/ decomposition — large skills auto-decompose for token budget awareness; detail loaded on demand, not upfront
5-platform portability — export skills as portable archives for Claude, Cursor, Codex, Copilot, Gemini; single authoring format, multiple deployment targets
MCP server with 4 tools for skill distribution — publish/install lifecycle for .tar.gz skill packages; package manager pattern for the skill ecosystem

Breaking Changes

None.