delphi-checkpoint
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell command to record project state.
- Evidence:
delphi checkpoint --note "$ARGUMENTS"inSKILL.md. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data that is interpolated into a shell command.
- Ingestion points: Untrusted data enters via the
$ARGUMENTSvariable inSKILL.md. - Boundary markers: The input is wrapped in double quotes (
"$ARGUMENTS") which provides basic shell delimitation but does not prevent all forms of command substitution. - Capability inventory: Executes the
delphiCLI tool with file access capabilities implied by the description ("records STATE as it stands, the git HEAD and branch"). - Sanitization: There is no explicit validation or sanitization logic within the skill to filter malicious shell characters or command sequences.
Audit Metadata