The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a shell command pipeline (ls, sort, tail) to determine the next sequence number for ADR files. The command uses a hardcoded path (docs/architecture/) and does not incorporate untrusted user input, making it safe.
[DATA_EXFILTRATION]: No network operations, external downloads, or access to sensitive credentials (such as .ssh or .aws directories) were detected. The skill's operations are confined to the local project documentation directory.
[PROMPT_INJECTION]: The instructions do not contain patterns designed to bypass safety filters, override agent behavior, or extract system prompts.
[INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes untrusted data from the local environment.
Ingestion points: Reads PRD files and implementation code as specified in SKILL.md to extract decisions.
Boundary markers: Absent. The skill does not explicitly use delimiters to wrap external content.
Capability inventory: File system write access to docs/architecture/ and shell command execution for file listing in SKILL.md.
Sanitization: Absent. Content is processed directly without specific filtering for malicious instructions.

adr