Ai

GitHub Spec Kit v0.7.5 Tightens Agent Safety and Adds Self-Check Tools

GitHub Spec Kit v0.7.5 Tightens Agent Safety and Adds Self-Check Tools

GitHub has released Spec Kit v0.7.5, a focused update that improves agent safety, expands extension support, and lays groundwork for easier CLI maintenance. While this is not a massive feature release, it includes meaningful changes for teams using SKILL.md-based agents, community extensions, and secure command execution workflows.

What Changed

The most important functional fix in v0.7.5 is broader SKILL.md placeholder resolution. Spec Kit now resolves skill placeholders for all SKILL.md agents rather than only codex and kimi, which should make multi-agent workflows more consistent across different agent types.

The release also introduces a new specify self check command along with a self-upgrade stub in the CLI. That suggests GitHub is moving toward better built-in diagnostics and lifecycle management for local Spec Kit installations, even if the upgrade path is still early-stage.

Another high-impact improvement is a security fix that blocks directory traversal in agent command write paths. This reduces the risk of unintended file writes outside expected directories and strengthens the safety model for agent-driven command execution.

On the ecosystem side, GitHub expanded the catalog with new community extensions, including red-team, superpowers-bridge, and a wireframe extension. It also reorganized documentation by moving community presets and walkthroughs out of the main README into dedicated docs/community sections, making the project structure cleaner as the catalog grows.

Why It Matters

For teams building structured AI workflows, the expanded SKILL.md placeholder support is the biggest usability gain in this version. It reduces agent-specific inconsistency and should make shared templates and skills more portable across different runtime setups.

The directory traversal fix is especially relevant for enterprise and developer-tooling audiences. As agent frameworks become more capable, filesystem safety becomes a core trust requirement. Hardening write paths is the kind of low-visibility change that matters a lot in real deployments.

The new self-check command is also worth watching. Even in stub form, it points toward a more mature CLI experience where developers can validate their local environment and potentially streamline upgrades without relying on manual troubleshooting.

Overall, v0.7.5 looks like a stabilization and quality release rather than a headline-grabbing platform shift. But for users already adopting Spec Kit, the combination of safer agent execution, broader skill handling, and a growing extension catalog makes this a practical update with clear operational value.

Official Source: https://github.com/github/spec-kit/releases/tag/v0.7.5