Big Memory
ReviewAudited by ClawScan on May 1, 2026.
Overview
Big Memory is a coherent instruction-only memory helper, but it persistently saves task and code context, so users should review its privacy implications before relying on it.
This skill appears consistent with its advertised memory-recovery purpose. Before installing, decide whether you are comfortable with automatic durable snapshots of code, file paths, decisions, and task state. Avoid saving secrets, review memory files periodically, and be cautious with optional session-memory or cloud-embedding settings.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may add task snapshots to memory logs without a separate save command when its triggers fire.
The skill directs the agent to create snapshots and modify local memory files, including some agent-initiated triggers. This is expected for a memory snapshot skill, but it is still local file mutation.
After completing a significant unit of work ... If yes, create a snapshot. ... APPEND the snapshot ... use `Edit` ... Alternatively, use `Write`
Confirm you are comfortable with append-only memory file writes, and review or prune the memory folder periodically.
Private implementation details or accidentally included secrets could remain in durable memory and be retrieved in later sessions.
The skill intentionally persists detailed task context, including code, paths, identifiers, and configuration-related values. This is purpose-aligned, but such memory can contain sensitive project information if users are not careful.
Captures exact code, decisions, file paths, and task state ... Key Names & Values ... Env var: `DATABASE_URL`
Do not include passwords, API keys, tokens, or secret environment values in snapshots; review saved memory files before sharing a project or enabling broad recall.
If enabled, past conversation content may become searchable, and with a cloud provider some memory/search content may be processed outside the local machine.
The optional configuration can expand retrieval to conversation transcripts and can use a cloud embedding provider. This is disclosed and optional, but it changes where session data may be indexed or processed.
`sessionMemory`: true ... `sources`: ["memory", "sessions"] ... If you prefer cloud embeddings, use `"provider": "openai"`
Use the local provider for sensitive projects, enable session memory only when needed, and check your organization’s data handling rules before using cloud embeddings.