The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script scripts/neon-soul.mjs using Node.js to perform the synthesis logic. It also provides instructions for establishing persistence by adding a scheduled task through the openclaw cron command.\n- [DATA_EXFILTRATION]: The skill accesses sensitive file paths including the user's memory/ directory and agent session logs at ~/.openclaw/agents/main/sessions/*.jsonl. This involves reading private interaction history and agent memory to extract behavioral patterns for identity synthesis.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from memory files and session logs that are subsequently processed by an LLM.\n
Ingestion points: Files in the memory/ directory and agent session logs at ~/.openclaw/agents/main/sessions/.\n
Boundary markers: No specific delimiters or instructions to ignore embedded commands were identified in the logic or documentation.\n
Capability inventory: The skill executes local shell commands via Node.js and has the ability to read and write to the local file system.\n
Sanitization: No validation, sanitization, or filtering of the memory or log content is described before it is passed to the synthesis engine.

consciousness-soul-identity