The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to invoke powershell.exe, executing a script to process 1C report files.
[DYNAMIC_EXECUTION]: The instructions direct the agent to execute a PowerShell script located in a sibling directory (${CLAUDE_SKILL_DIR}/../epf-dump/scripts/epf-dump.ps1). This represents a dependency on local external code via relative path traversal.
[CREDENTIALS_UNSAFE]: The skill provides parameters for -UserName and -Password to be passed directly as command-line arguments to the PowerShell script. This practice can result in credentials being exposed in process listings or shell history.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external files to determine command arguments, creating a potential surface for indirect injection.
Ingestion points: Reads the .v8-project.json file from the project root to resolve database paths, aliases, and branch mappings.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are provided for the content read from the JSON configuration.
Capability inventory: The skill can execute shell commands via Bash, read files, and access directory structures using Glob and Read tools.
Sanitization: There are no explicit instructions to sanitize or validate the strings retrieved from .v8-project.json before they are interpolated into the shell command execution.

erf-dump