daily-recap
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes historical session logs containing untrusted user input to generate daily reports.
- Ingestion points: In Step 3, researcher subagents read the opening and closing messages of significant sessions identified in the session list.
- Boundary markers: The instructions do not specify any delimiters or safety warnings to ensure the subagents ignore instructions embedded within the session data being scanned.
- Capability inventory: The skill uses
mkdir -pfor directory creation and writes summarized content to files in the~/.pi/recap/directory. It also spawns subagents to process the data. - Sanitization: No sanitization, escaping, or validation of the session content is described before it is passed to the subagents for scanning.
- [COMMAND_EXECUTION]: The skill uses shell commands for file management and metadata extraction.
- Uses
date +%Vto retrieve the ISO week number for file path construction. - Uses
mkdir -pto create recursive directory structures for the recap files. - Uses
stat -c %Yto retrieve file modification timestamps for session filtering and triage.
Audit Metadata