EngramClaw

v1.0.0

Sistema de memoria persistente para agentes IA. Usa mem_save después de bugfixes, decisiones, descubrimientos, cambios de config. Usa mem_search cuando el us...

0· 223·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
Name/description (persistent memory for agents) match the declared needs: two CLI binaries (mcporter + engram) and MCP tools invoked via mcporter. Requiring mcporter to expose Engram's MCP tools is consistent with the described integration.
Instruction Scope
SKILL.md instructs the agent when to call mem_save/mem_search/mem_session_summary and explicitly encourages the agent to 'decide proactively' what is significant and save it. The instructions do not tell the agent to read arbitrary system files or request unrelated credentials, but they do permit storing session summaries and references to project files. There is no guidance about redacting secrets or restricting what gets saved, so the agent could end up persisting sensitive data if not constrained.
Install Mechanism
Install steps use Homebrew formulas and npm for mcporter, and GitHub Releases/manual binaries for Engram. These are standard distribution mechanisms; no obscure download shorteners or unknown servers are used in the instructions. The manual binary install writes an executable to /usr/local/bin which is expected for CLI tools.
Credentials
The skill requests no environment variables or external credentials, which is proportionate to a local CLI-based tool. However, the SKILL.md does not document where Engram stores its data, what access controls or encryption are applied, or whether the storage might be networked — this is an important omission because the tool will persist session data.
Persistence & Privilege
always:false and user-invocable:true. The skill can be invoked autonomously (platform default) but it does not demand permanent or system-wide privileges, nor does it attempt to modify other skills. Combine this with the instruction to proactively save important findings: autonomous agent behavior could persist data without explicit user prompts, so consider limiting autonomous calls if that is a concern.
Assessment
This skill appears to do what it says: expose a local/personal memory backend through MCPporter. Before installing, review and decide on these points: 1) Where does Engram store its data (local file path, DB)? Ensure it's not stored in an unsecured or networked location you didn't expect. 2) Does Engram encrypt or restrict access to its storage, and can you configure retention or automatic redaction? If not, avoid saving secrets. 3) Because the SKILL.md tells the agent to 'decide proactively' to mem_save, consider preventing autonomous invocation or restricting the agent's permission to call tools until you trust its behavior. 4) Verify the Homebrew formulas / GitHub Releases checksums and audit the Engram repository (or run it in an isolated environment) if you will store sensitive project info. 5) Prefer explicit mem_save calls triggered by you rather than letting the agent save automatically until you are confident with its filtering and storage controls.

Like a lobster shell, security has layers — review code before you run it.

latestvk9774kkhpa1g9areyqm574xvcd82msgr

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Runtime requirements

🧠 Clawdis
Binsmcporter, engram

Install

Install MCPorter via Homebrew (macOS/Linux)
Bins: mcporter
brew install steipete/tap/mcporter
Install Engram via Homebrew (macOS/Linux)
Bins: engram
brew install gentleman-programming/tap/engram

Comments