SwarmRecall
ReviewAudited by ClawScan on May 10, 2026.
Overview
SwarmRecall is a coherent persistent-memory integration, but it will store and reuse agent data on SwarmRecall servers using an API key, with optional sharing and background consolidation.
Install this if you want a cloud-backed persistent memory service for your agent. Before use, decide what may be stored externally, avoid secrets or sensitive logs, use shared pools only when intended, keep the API key private, and review the npm CLI/source for high-sensitivity deployments.
Findings (5)
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.
Information the agent stores may be retained, searched, summarized, and used in future responses.
The core feature is persistent external storage and reuse of agent context, which is expected for this skill but can preserve sensitive or incorrect context across sessions.
Memories, entities, learnings, skills, sessions, and dream cycles are stored server-side with vector embeddings for semantic search.
Only store information the user has agreed to persist externally; avoid secrets and raw logs with credentials; periodically review and delete obsolete or incorrect memories.
Anyone or any process with the API key may be able to access or modify the agent's SwarmRecall data.
The integration uses a local/API credential to authenticate agent access to SwarmRecall data.
This creates `~/.config/swarmrecall/config.json` with your API key
Use a dedicated key per agent, keep it out of shared files and source control, and revoke/rotate it if exposed.
Data written to a shared pool may be seen by other agents or users with pool access.
The skill supports shared collaboration pools, so stored memories or graph data may become visible to other pool members when a pool is used.
Pass `poolId` to any create call to write into a shared pool. Pool members will see your writes
Use private storage for personal or confidential content, and verify pool membership/access before writing to a pool.
If enabled, background consolidation may continue processing stored memories after the immediate task.
Dream consolidation can be configured for recurring/background processing and can prune or clean up stored data; this is disclosed and tied to the product purpose.
`swarmrecall dream config [--enable] [--disable] [--interval <h>]` ... `swarmrecall dream execute [--ops <o>]` | Run Tier 1 ops (decay, prune, cleanup).
Enable dream cycles only when desired, use dry-run/review options where available, and confirm which operations may prune or alter stored data.
Installing the skill means trusting the published npm package to implement the documented behavior.
The skill relies on an external npm CLI/MCP server rather than bundled reviewed runtime code.
node | package: @swarmrecall/cli | creates binaries: swarmrecall
Install from the expected npm package, consider pinning a reviewed version, and inspect the package/source before using it in sensitive environments.