Editorial policy

We prioritize tools relevant to AI coding agents, MCP servers, agent skills, repositories, and adjacent developer workflows. Public submissions may inform coverage, but submission does not guarantee review, publication, placement, or a particular score.

• A listing may be declined or removed when information is too limited, misleading, unsafe to present without context, outside the directory's scope, or no longer useful.
• Featured placement is an editorial label within the directory and should not be read as a warranty.

Writers should distinguish observable repository facts from maintainer claims and editorial interpretation. Material statements should be traceable to current public sources where practical. We avoid presenting estimates, generated summaries, or compatibility claims as independently tested facts when they are not.

• Check the public project identity, repository, documentation, license signal, maintenance activity, setup path, and relevant risk notes.
• Write for readers who need to understand the tool quickly without removing important caveats.
• Use conservative language where evidence is incomplete or could change.
• Revisit content when a material correction or meaningful public change is identified.

A submission, repository popularity, maintainer relationship, or potential commercial interest should not determine factual conclusions. If sponsored or paid placement is introduced, it should be clearly labeled rather than blended into ordinary editorial coverage.

Automation may help collect public metadata or draft structured information. Automated output can be incomplete or wrong. The AgentHubStack Editorial Team remains responsible for published wording and should apply human review before publication.

We correct material factual errors when we can verify them. Disagreement with an opinion or score does not by itself establish an error, but new public evidence may justify a review.