The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data retrieved from external tools, creating a potential surface for indirect prompt injection.
Ingestion points: Metric records are fetched via mcp__plugin_ruflo-core_ruflo__memory_search in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potentially malicious content embedded within the retrieved metrics.
Capability inventory: The skill utilizes various memory and pattern storage tools (memory_*, agentdb_*) and has Bash access listed in its allowed-tools configuration.
Sanitization: No explicit sanitization or validation steps are defined for the data retrieved from the observability namespace before it is aggregated or reported.
[EXTERNAL_DOWNLOADS]: The skill's documentation references the execution of external code from a public registry.
Evidence: The 'CLI alternative' section suggests the use of npx @claude-flow/cli@latest, which downloads and executes a package from the npm registry at runtime.

observe-metrics