The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses the directory ~/.claude/session-transcripts/ to read raw session logs. While these files contain sensitive historical interaction data, the access is localized and necessary for the core functionality of memory consolidation.
[COMMAND_EXECUTION]: The instructions include the use of shell commands such as grep to process and search through local transcript files.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing untrusted data from session logs which could contain malicious instructions meant to persist in long-term memory.
Ingestion points: Reads transcripts from ~/.claude/session-transcripts/ as described in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified in the consolidation workflow.
Capability inventory: Utilizes grep for file reading and performs writes to MEMORY.md and supplementary topic files.
Sanitization: There is no mention of filtering, escaping, or validating the content extracted from logs before it is saved to memory files.

rem-sleep