nyldn/claude-octopus v9.0.0

v9.0.0 — CC v2.1.76 sync, claude-mem integration, command audit

on GitHub

What's New in v9.0.0

CC v2.1.76 Feature Sync

• 6 new SUPPORTS_* detection flags (100 total, 31 version_compare blocks)
• MCP elicitation, elicitation hooks, worktree sparse paths, post-compact hook, effort command, BG partial results
• 5 new /octo:doctor checks wired

claude-mem Companion Integration

• Auto-detects claude-mem plugin at session start
• scripts/claude-mem-bridge.sh — bidirectional integration (search, observe, context)
• Phase completion observations written to claude-mem (non-blocking)
• 6 skill files with MCP tool hints for cross-session memory

Command Audit Remediation (7 commands fixed)

• /octo:staged-review — rewritten: broken /octo:verify + /octo:ship refs fixed, compliance block, AskUserQuestion scope, interactive next steps
• /octo:review — "All areas (Recommended)" 5th focus option
• /octo:brainstorm — Solo/Team mode with multi-provider dispatch
• /octo:prd — parallel multi-provider Phase 1 research
• /octo:prd-score — optional "Rigorous" multi-AI consensus scoring
• /octo:model-config — stale GPT-5.3-Codex-Spark → GPT-5.4
• 3 new test suites (28 tests): meta-prompt, prd-score, staged-review

Other

• Path traversal fix in loadCommands/loadSkills (closes #150)
• Provider report card + persistent fallback log for /octo:doctor
• /octo:review UX — default all focus areas, auth preflight

Stats

• 32 personas, 39 commands, 50 skills
• 100 feature detection flags, 31 version_compare blocks
• 81 tests passing (pre-push suite)

Full Changelog: v8.57.0...v9.0.0