The Agent Skills Directory

[SAFE]: The skill instructions and templates define a benign workflow for project documentation and context management within the local filesystem. No obfuscation, prompt injection, or suspicious commands are present.- [NO_CODE]: The skill consists exclusively of Markdown templates and natural language instructions. It does not include executable scripts, shell commands, or external binary dependencies.- [DATA_EXPOSURE]: The skill facilitates writing session-specific metadata (file paths, decision logs, progress) to the local memory-bank/ folder. This data remains on the local machine and is used for productivity; no network exfiltration patterns were identified.- [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by reading from and writing to local Markdown files that may be influenced by external project data.
Ingestion points: Reads from files in the memory-bank/ directory, such as activeContext.md and progress.md.
Boundary markers: The templates do not currently implement specific delimiters or 'ignore' instructions for embedded content.
Capability inventory: The skill performs local file-write operations to maintain the memory bank.
Sanitization: No explicit sanitization or validation logic is defined for the content stored in these files.

memory-checkpoint