The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill features a data ingestion surface as it reads local source code and existing documentation to map the current system architecture. This is a primary function of the skill to ensure specifications are accurate.
Ingestion points: Reads local source code files (.ts, .py, config, schema) and markdown documentation in proposals/, decisions/, specs/, and guides/ directories (Step 1).
Boundary markers: The instructions do not specify the use of delimiters when interpolating code content into the specification draft.
Capability inventory: The skill utilizes tools to list directories (exec), read files (exec), and write new markdown files (write tool).
Sanitization: No explicit sanitization or filtering of the ingested source code is performed before inclusion in the draft spec.
[COMMAND_EXECUTION]: The skill uses an exec tool to perform directory listings (ls -A) and read file contents (cat). These actions are restricted to specific project subdirectories (proposals/, decisions/, specs/) to gather context for the specification being written.
[EXTERNAL_DOWNLOADS]: The skill metadata references the author's official GitHub repository (github.com/inkeep/open-knowledge-skills) as the source for the software-lifecycle pack. This is a standard vendor-provided resource for the skill's functionality.

write-a-spec