Self-orchestrating multi-agent development workflows. You say WHAT, the AI decides HOW.
v5.11.3 – Security Declaration Alignment This release fixes security-scan findings caused by inconsistent runtime declarations in v5.11.2. What changed • Updated runtime declarations to match actual workflow behavior: • requires_binaries: true • requires_credentials: true • requires_network: true • Clarified package behavior: • The package itself remains documentation-only (no install-time executable payload). • Agent workflows may execute shell/network actions at runtime when invoked by the user. • Added explicit runtime transparency: • Required/optional runtime binaries • Optional credentials/environment expectations (e.g. GH_TOKEN, Claude auth, MCP configuration) • Refined security semantics to distinguish: • install-time safety • runtime agent capabilities Why this release v5.11.2 could create a misleading security impression by claiming no network/credentials/code-execution requirements while documenting workflows that use shell, GitHub, Playwright, and web tools. v5.11.3 resolves that mismatch for clearer trust and more accurate scanner interpretation. Compatibility • No breaking workflow changes • Backward-compatible behavior • Metadata/security declaration update only