The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes bash commands to manage session state. Specifically, it retrieves the current session ID via echo $CLAUDE_SESSION_ID, generates unique fallback identifiers by reading the first 8 characters of /proc/sys/kernel/random/uuid, and captures project activity using git diff --name-only HEAD.
[DATA_EXPOSURE]: The skill maintains session persistence by reading and writing JSON state files within the system's temporary directory (/tmp/workflow-session-*.json). These files are used to store and retrieve session slugs.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests untrusted data from multiple sources:
Ingestion points: User-provided project titles, descriptions, and session summaries are collected via interactive prompts.
Boundary markers: No explicit delimiters or instructions are used to prevent the agent from obeying instructions embedded in the user-provided text.
Capability inventory: The skill can execute shell commands (bash) and interact with file system paths in /tmp.
Sanitization: There is no evidence of sanitization or escaping of the user-provided strings before they are passed to the workflow_session_end or workflow_project_create tools.

workflow