hivecontrolagents/stingward is an inspection tool that finds security risks in AI-built apps, Claude Code skills and Model Context Protocol (MCP) servers. The latest release recorded in the repository is v0.1.6. The provided material contains no changelog for this version, so no newly added or changed features can be confirmed.
The tool states that it checks for vulnerable dependencies, secret exposure, risky agent configuration and code flows, and hidden instructions that only the model can read. Its examples include exposure of environment variables such as a JavaScript public key and overly broad shell execution permissions. It can run in terminals, as a developer tool extension and in code repository workflows, and it provides guidance for Python and JavaScript analysis styles. There are also examples that show inspection results graded from A to F. The repository's primary language is Python and the license is Apache-2.0.
The verifiable evidence, however, is limited. The repository creation timestamp is 2026-10-19T22:50:37Z and the star count is 2. The research record's momentum score is 1.2, and no novelty, technique or quality scores were provided. The grading criteria are also absent from the given material. It is therefore difficult to judge actual detection performance from the grades and momentum score alone; for now, it is more reasonable to treat this as a security tool worth reviewing for the scope of what it checks.