The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses the ! syntax in the Project Context section to execute git log and ls commands at load time. This is used to orient the agent to the current state of the development project. These are benign, standard development commands and do not involve network exfiltration or access to sensitive configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it ingests data from commit messages and filenames which could be controlled by an external party.
Ingestion points: SKILL.md uses the output of git log and ls to populate the agent's context.
Boundary markers: None are present to delimit the command output from the instructions.
Capability inventory: The skill is primarily instructional, guiding the agent to use other specialized tools for building, testing, and reviewing code.
Sanitization: No sanitization or filtering is applied to the output of the context-gathering commands.

app-development-workflow