The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The library script hooks/lib/workspace-detector.sh utilizes the eval command to perform indirect expansion of environment variables (eval echo "\${$api_key_env:-}"). This involves executing a string as a command, which can lead to arbitrary code execution if the configuration source providing the variable name is compromised.
[COMMAND_EXECUTION]: The skill implements automated lifecycle hooks (hooks/pre-prompt.sh and hooks/stop.sh) that execute shell commands on the host system. These scripts run automatically before and after agent sessions, performing operations with git, curl, and jq to manage board state and session logs.
[DATA_EXFILTRATION]: At the end of a session, the hooks/stop.sh script automatically aggregates sensitive repository data, including staged git diffs, commit messages, and a list of modified files. This information is transmitted to external Plane API endpoints (e.g., plane.delo.sh or plane.internal.intelliforia.com).
[DATA_EXPOSURE]: The skill processes untrusted data from external sources, including Plane API responses (ticket titles and descriptions) and local git metadata (branch names and commit history). This creates an attack surface for indirect prompt injection where malicious content in a ticket could influence agent behavior during processing.
[DATA_EXPOSURE]: The skill reads workspace configuration and project identifiers from local .plane.json files and a global configuration at ~/.claude/plane-workspaces.json to resolve API keys and endpoint URLs.

managing-tickets-and-tasks-in-plane