The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from MOOS log files, creating a potential surface for instructions embedded in data to influence the agent.
Ingestion points: Data is read from .alog files using tools such as alogscan, aloggrep, and aloghelm.
Boundary markers: Absent. The skill provides no instructions to use delimiters or ignore potential commands within the log content.
Capability inventory: The agent is authorized to execute specific binaries (alog*), a local wrapper script (scripts/alogvars.sh), and generic shell/Python code for analysis tasks.
Sanitization: Absent. There is no evidence of content sanitization or validation before processing.
[COMMAND_EXECUTION]: The skill relies on executing system-level binaries and local scripts to perform its primary function.
It assumes the availability of aloggrep, aloghelm, and alogscan in the path.
It uses a custom bash script scripts/alogvars.sh which executes alogscan and processes the results via awk.
[DYNAMIC_EXECUTION]: The instructions explicitly direct the agent to generate and run shell or Python code for custom data analysis.
Evidence: Instructions in SKILL.md and references/alog-tool-guide.md specify using "shell or Python for custom logic" for geometric or numeric questions.

moos-alog-analysis