piclist-upload
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/process.shscript executes shell commands includingcurl,rm,mv,find,awk,sed,grep, andjq. These are standard utilities used to implement the skill's core functionality of processing Markdown files and communicating with the PicList HTTP server. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from Markdown files which could potentially contain malicious instructions if the files originate from untrusted sources.
- Ingestion points: Local Markdown files and their contained image paths are read and parsed by
scripts/process.sh. - Boundary markers: The script lacks explicit markers to differentiate content from instructions but employs specific regex patterns to target image references only.
- Capability inventory: The skill possesses network upload capabilities via
curland file system modification/deletion capabilities viarmandmv. - Sanitization: Path normalization and percent-encoding decoding are applied to extracted image paths to ensure they resolve to valid local files before processing.
- [SAFE]: A static analysis flag for a reverse shell pattern in
scripts/process.shwas determined to be a false positive. The code(exec 3<>"/dev/tcp/127.0.0.1/$port")is a standard Bash mechanism used to verify the availability of the local PicList server port and does not facilitate unauthorized remote access.
Audit Metadata