heating the forge...

@forgecat/awslabs_aidlc-workflows — Security Risk Report | ForgeCat

awslabs_aidlc-workflows

Security risk report for @forgecat/awslabs_aidlc-workflows v0.0.7

Low

Profile sourced from AWS Labs official repository (https://github.com/awslabs/aidlc-workflows) with documented commit hash and MIT-0 license.
No typosquatted or attacker-controlled package dependencies declared.
Static scanner confirms safe level.

Low

Content describes a legitimate software development workflow framework (AI-DLC) with structured planning, requirements gathering, and design phases.
No instructions to ignore system prompts, read credentials, exfiltrate data, or install remote payloads.
Guidance on question formats, error handling, and session continuity is standard practice documentation, not poisoning—recommends well-known legitimate practices (Mermaid diagrams, ASCII standards, markdown validation).

Details

Findings

Invisible characters detected
rules/aws-aidlc-rules/core-workflow.md · SKILL-011
Invisible characters detected
rules/aws-aidlc-rule-details/common/welcome-message.md · SKILL-011
Invisible characters detected
rules/aws-aidlc-rule-details/inception/application-design.md · SKILL-011

Evidence

README.md: 'Converted rule bundle for teams using AI-DLC core workflow with detailed stage rules.'
welcome-message.md: 'I'll guide you through an adaptive software development workflow that intelligently tailors itself to your specific needs.'
overconfidence-prevention.md: 'Default to Asking: When there's any ambiguity, ask clarifying questions' — standard best practice, not malicious guidance.

Low

Profile declares no tools, skills, or MCP servers—it is a rule-only bundle (markdown guidance documents).
No file_write, file_delete, shell, or web_fetch permissions requested or embedded.
Content provides guidance on question formats and workflow stages but does not demand execution authority beyond reading and analyzing user responses.

Details

Evidence

README.md: 'Skills: None (rule-only profile)'
No tool definitions, MCP servers, or alwaysApply rules in profile content.
All directives are instructional (e.g., 'Create question file', 'Read artifact') without claiming autonomous execution rights.

Low

Details

Evidence

Profile metadata: 'MCP servers (untrusted data): (none)'
All content is static markdown documentation without server definitions or network calls.