The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The install.sh script clones the official awslabs/aidlc-workflows repository from GitHub to retrieve the core rules and templates required for the skill's operation. This is a neutral operation fetching resources from a well-known service.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted user input via the $ARGUMENTS parameter.
Ingestion points: User-provided task descriptions are interpolated into the prompt in SKILL.md.
Boundary markers: The skill uses clear markdown section headers to separate instructions from data, though specific "ignore instructions" delimiters are not explicitly defined in the top-level file.
Capability inventory: The skill utilizes powerful tools including Bash, Write, Edit, and Task to modify code and execute shell commands.
Sanitization: Risk is mitigated by a mandatory "Wait for Approval" protocol, where the agent must present an execution plan and obtain explicit user consent before performing actions.
[COMMAND_EXECUTION]: The skill uses Bash and Task tools to perform standard software development lifecycle activities such as workspace detection, build execution, and testing. These operations are consistent with the skill's stated purpose.
[DATA_EXFILTRATION]: The skill maintains project state and interaction logs within the local aidlc-docs/ directory. No network operations were found that exfiltrate this data to external or non-whitelisted domains.

aidlc