The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses node -e to execute an inline JavaScript snippet that reads and parses the ~/.claude-mem/settings.json file to resolve network ports.
[COMMAND_EXECUTION]: The skill utilizes the sqlite3 CLI tool to perform multiple direct queries against the local database located at ~/.claude-mem/claude-mem.db.
[CREDENTIALS_UNSAFE]: The skill accesses files within the user's home directory (~/.claude-mem/) which contain persistent memory logs and configuration data. These files are treated as sensitive as they represent the user's long-term interaction history.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted historical data (the project timeline) which could contain malicious instructions from previous sessions intended to influence the subagent.
Ingestion points: The full compressed timeline is fetched from a local worker API and passed directly as context to a subagent in Step 4.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the subagent's prompt template.
Capability inventory: The subagent is explicitly instructed to execute sqlite3 commands, providing a path to command execution if the input data is compromised.
Sanitization: The skill lacks sanitization or filtering for the historical timeline data before it is processed by the subagent.

timeline-report