The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: Downloads and executes the @claude-flow/cli package from the NPM registry using the npx command. This tool is part of the project's ecosystem and is used for memory transfer operations.
[COMMAND_EXECUTION]: Utilizes the Bash tool to execute shell commands for listing, exporting, and importing memory patterns.
[PROMPT_INJECTION]: Identifies a surface for indirect prompt injection because the skill ingests external data from RVF memory files.
Ingestion points: Data is brought into the agent context via the mcp__claude-flow__memory_import_claude tool and the hooks transfer from-project CLI command.
Boundary markers: The skill does not implement delimiters or specific instructions to ignore malicious prompts that might be embedded within the imported memory data.
Capability inventory: The agent has access to the Bash tool and multiple memory management functions, which could be leveraged if malicious instructions are successfully injected.
Sanitization: There is no evidence of validation or sanitization of the content within the imported RVF files before it is processed by the agent.
[REMEDIATION]: To mitigate risks, the agent should treat imported memory as untrusted data. It is recommended to use clear boundary markers when processing this content and to implement a review step or sanitization process for any instructions found within the imported metadata or vector entries.

rvf-manage