The Agent Skills Directory

[SAFE]: The skill manages local CRM data by reading and writing to markdown files such as crm/outreach.md and crm/companies.md. This activity is consistent with its stated purpose of manual outreach logging and does not involve any external network communication.
[COMMAND_EXECUTION]: The skill utilizes Bash(test *) and Bash(date *) to verify the existence of files and retrieve current dates. These commands are limited in scope and do not pose a risk of privilege escalation or arbitrary code execution.
[SAFE]: Evaluation of the attack surface for indirect prompt injection:
Ingestion points: User-provided strings from $ARGUMENTS (such as company names and notes) are processed by the skill in SKILL.md.
Boundary markers: There are no explicit delimiters defined to separate user input from system instructions, although the output is constrained to a structured markdown table format.
Capability inventory: The skill utilizes Read, Edit, Write, and restricted Bash tools to manage local files.
Sanitization: The skill implements kebab-case normalization when generating company slugs from user input, providing a basic level of validation for those strings.

log-outreach