The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard system utilities including find, unzip, sed, and ripgrep (rg) to locate, extract, and search through diagnostic files and source code. It also leverages the official Eclipse Memory Analyzer (MAT) command-line interface (ParseHeapDump.sh) to process large heap dump files. All commands are targeted at local files for the purpose of technical analysis.
[DYNAMIC_EXECUTION]: A hardcoded Python script is used as a utility to clean up and extract plain text from generated HTML reports. The script uses standard libraries for regex and HTML parsing to strip tags and script blocks. It does not ingest untrusted data as executable code and functions as a safe text-processing helper.
[INDIRECT_PROMPT_INJECTION]: The agent is instructed to read and summarize content from externally generated reports (MAT HTML files). While this constitutes an ingestion point for untrusted data, the skill provides a specific context for technical analysis (looking for retained heap evidence and class names), which inherently limits the effectiveness of typical natural language injection attempts within this workflow.

java-heap-dump-triage