The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local shell commands (sed, grep, tr, cut) to normalize user input into feature slugs and directory names. It also leverages git for branch creation and committing specification artifacts. These operations are essential to the skill's functionality and are performed locally without escalation.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests and processes untrusted data from user feature requests and external roadmap files.
Ingestion points: The $ARGUMENTS variable containing the user's feature description and existing entries in .spec-flow/memory/roadmap.md.
Boundary markers: No explicit delimiters are implemented to separate external content from system instructions in the specification prompt.
Capability inventory: The skill possesses the capability to write to the local filesystem (spec.md, state.yaml) and interact with the git repository.
Sanitization: User input used for slugs and filenames is processed via a restrictive regex ([^a-z0-9-]), which strips potentially malicious characters and shell control sequences, significantly mitigating common injection vectors.

specification-phase