v15.8.0 "Governed Integrations and Agent Project Workflows"

[15.8.0] - 2026-08-02 - "Governed Integrations and Agent Project Workflows"

Added four evidence-first skills for GitHub attachments, governed AI gateways, project-specific agent instructions, and revenue analytics through MCP.

Start here:

• AAS Core setup: configure the exact aas runtime with the Core guide
• Direct skill distribution: npx agentic-awesome-skills
• Choose your tool
• Best skills by tool
• Bundles
• Workflows

Added

• Added gh-attach for uploading and downloading GitHub user attachments through the gh CLI, with repository-scoped visibility and GitHub Enterprise Server support.
• Added unified-ai-gateway for governed inspection of Unified AI System's eight Codex MCP tools, with pinned image provenance, non-executing review, and separate activation approvals.
• Added agents-generator for generating project-specific AGENTS.md and companion rules with package-manager detection, monorepo handling, backups, dry-run/update modes, and command validation.
• Added talivia-agent-kit for revenue-first website analytics through the official Talivia MCP server, with explicit confirmation for tracking, deployment, and payment-attribution changes.

Changed

• Regenerated the canonical catalog and Codex/Claude distribution surfaces for 2,001 skills.

Who should care

• Developers who need to attach or retrieve GitHub artifacts directly from the terminal.
• Teams evaluating a self-hosted AI gateway with bounded, fake-provider, and approval-aware MCP workflows.
• Maintainers who want reproducible project-specific agent instructions and companion rules.
• Revenue and growth teams using Talivia analytics while keeping website, deployment, and payment-attribution mutations consent-gated.

Validation

• Passed canonical skill and reference validation, documentation security, warning-budget enforcement, repository tests, protected CI, and CodeQL.
• Confirmed canonical synchronization reproduced the four new skills across Codex and Claude plugin distributions with no generated-state drift.
• Confirmed npm package audit has no moderate-or-higher production vulnerabilities on the release base.

Credits

• sudosubin/gh-attach for gh-attach in PR #1060.
• happy520ai/unified-ai-system for unified-ai-gateway in PR #1061.
• OJPalenzuela/agents-generator for agents-generator in PR #1067, which closes issue #1065.
• talivia-group/agent for talivia-agent-kit in PR #1069.