The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to perform metadata analysis and content inspection of local files.
Evidence: Uses wc -l to count lines in context files at ~/.claude/.context/core/*.md.
Evidence: Uses head and grep to verify synchronization status in ~/.claude/rules/elle-core.md.
Evidence: Uses find to locate platform capability reference files within the plugin cache.
[INDIRECT_PROMPT_INJECTION]: The skill reads and summarizes content from multiple context files that could contain instructions embedded in data, which the agent might inadvertently follow.
Ingestion points: Reads multiple markdown files from ~/.claude/.context/core/ (e.g., identity.md, preferences.md, rules.md, relationships.md, triggers.md, workflows.md, session.md, projects.md, improvements.md).
Boundary markers: None; the instructions do not specify delimiters or provide warnings to the agent to ignore instructions embedded within the audit targets.
Capability inventory: The skill utilizes read-only shell tools (wc, head, grep, find) and file-reading capabilities. No network exfiltration or file-writing tools are invoked.
Sanitization: No evidence of sanitization or content filtering was found for the data ingested from the context files.

context-health