The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads and processes task files from kanban/tasks/ to generate a central board snapshot in kanban/BOARD.md.\n
Ingestion points: Task descriptions and metadata in kanban/tasks/*.md are ingested into the agent context during board generation and task listing operations.\n
Boundary markers: YAML frontmatter delimiters (---) are used for metadata isolation, but task bodies are unrestricted and could contain instructions.\n
Capability inventory: The skill can read and write files, manage directory structures, and execute its own management script within the repository.\n
Sanitization: Implements basic escaping for sed control characters when rendering task templates.\n- [DYNAMIC_EXECUTION]: The scripts/kanban utility dynamically loads shell variables from a repository-local configuration file using the source builtin.\n
Evidence: The script executes source "$config_file" to load settings from kanban/config.env. While this is a standard configuration pattern, it allows for the execution of shell commands if the configuration file is modified by an untrusted source.

local-kanban-board