The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill implements a '/money-skillify' mechanism that allows the agent to codify successful workflows into project-local skills. These are stored in a computed path at '~/.smtm/projects//skills/' and are automatically loaded into the agent's working context in subsequent sessions. This dynamic loading of locally generated content presents a risk if the generation process is influenced by malicious input.
[INDIRECT_PROMPT_INJECTION]: The skill performs 'Auto-Research' by scraping user-provided website URLs and searching for social profiles (LinkedIn, X, GitHub). The ingested data is used to build a 'User Profile' context block passed to all sub-skills. The instructions do not specify the use of boundary markers (delimiters) or sanitization techniques to ensure the agent disregards instructions that might be embedded in these external sources.
[COMMAND_EXECUTION]: The skill requires the agent to execute shell commands for telemetry logging (Step 2 of the Standard Skill Startup) and for performing an auto-update check using 'npm view' and 'cat' commands on local version files.
[EXTERNAL_DOWNLOADS]: The skill makes network calls to the NPM registry to check for package updates. It also references an external API service ('ccapi.ai') for AI model access. These references target well-known registries or the author's own infrastructure and are disclosed in the documentation.

money