ae-kb-bootstrap
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local bash scripts
scripts/company-scaffold.shandscripts/scaffold.shto perform directory scaffolding and template copying. These scripts use standard utilities likemkdir,cp,cat,awk, andsedto manage a local knowledge base structure. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (report definitions) via the
ae-cli analysis report getcommand. - Ingestion points: Data enters the context from the output of report and dashboard definition probes in
references/onboarding-flow.md. - Boundary markers: The workflow uses structured YAML and Markdown for the registry files and enforces 'hard gates' (confirmation steps 1, 2, and 3) where the agent must list results and wait for user approval before proceeding.
- Capability inventory: The skill can execute shell scripts, write to the local filesystem, and make network requests via the
ae-clitool. - Sanitization: Content is interpolated into templates using literal string replacement (via
awkinscaffold.sh), which lacks sophisticated escaping for potentially malicious prompt content embedded in field names or report metadata. - [DYNAMIC_EXECUTION]: The
scaffold.shscript dynamically generates Markdown and YAML registry files by injecting project-specific identifiers and probed metric calibers into pre-defined templates.
Audit Metadata