The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to silently execute deepvista upgrade check upon loading to verify the CLI version.
[COMMAND_EXECUTION]: The deepvista agents register command modifies the agent's platform configuration (e.g., ~/.claude/settings.json) to install a persistent "Stop" hook. This hook automatically executes status synchronization commands after every conversation turn to provide heartbeats to the vendor's dashboard.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation and upgrading of the deepvista-cli tool and its dependencies from public registries like PyPI and via the uv tool.
[DATA_EXFILTRATION]: The CLI manages sensitive authentication credentials stored in ~/.config/deepvista/credentials.json and transmits user-supplied knowledge base content, including notes and local files, to the external API at app.deepvista.ai.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted data from local files and notes for entity extraction, summarization, and the automated creation of new agent skills. The process lacks explicit boundary markers or content sanitization to prevent embedded instructions from influencing agent behavior.
[COMMAND_EXECUTION]: The skill includes functionality to generate and write executable SKILL.md files to local agent directories (e.g., ~/.agents/skills/), which constitutes dynamic script generation.

deepvista