ClawHub

Engram Memory

v0.3.1

Persistent long-term memory powered by knowledge graphs, ACT-R activation, and 15-phase consolidation. Remember conversations across sessions, recall relevan...

⭐ 1· 123·0 current·0 all-time
by@moshik21
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name and description describe a local memory/knowledge-graph system. Requiring either 'uv' (its package/tool runner) or Docker to run the server is proportionate to the stated purpose. Optional ANTHROPIC_API_KEY for richer entity extraction matches the description.
ℹ
Instruction Scope
SKILL.md instructs the agent to call local HTTP endpoints (127.0.0.1:8100) for observe/remember/recall/route and to optionally bootstrap with an absolute project_path. Those calls are within the skill's scope, but supplying a project_path or allowing the skill to call artifact/runtime endpoints lets the service access project files — users should understand this may expose local project data to the memory service.
ℹ
Install Mechanism
The registry entry is instruction-only (no automated install). SKILL.md directs standard installs: 'uv tool install engram', Docker image from ghcr.io, or cloning the GitHub repo. It also shows a curl of an install.sh from raw.githubusercontent.com; that is a well-known host but running downloaded scripts is higher-risk — the doc does advise reviewing the script before running.
✓
Credentials
No required environment variables. Optional 'ANTHROPIC_API_KEY' and 'ENGRAM_GROUP_ID' are reasonable and justified by the stated functionality. No unrelated credentials or broad secret requests are present.
✓
Persistence & Privilege
The skill is not set to always:true and does not request persistent platform-wide privileges or modify other skills. It is user-invocable and can be invoked autonomously (platform default), which aligns with its purpose.
Assessment
This skill expects you to run a local Engram server (Docker or uv-installed) that the agent will call on http://127.0.0.1:8100. Before installing or running anything: review the project's repository and any install scripts (the SKILL.md suggests 'less install.sh' before running), verify the Docker image source (ghcr.io/moshik21/engram), and only provide ANTHROPIC_API_KEY if you trust external API calls. Be aware that giving a project_path or enabling artifact/runtime routing allows the memory service to read project files — only provide paths you are comfortable sharing with the local service. If you need higher assurance, run the service in an isolated environment or inspect the source code fully before installation.

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

cognitive-architecturevk9784jjqfw5mgawgggbvbc693h833znzknowledge-graphvk9784jjqfw5mgawgggbvbc693h833znzlatestvk9784jjqfw5mgawgggbvbc693h833znzlong-term-memoryvk9784jjqfw5mgawgggbvbc693h833znzmcpvk9784jjqfw5mgawgggbvbc693h833znzmemoryvk9784jjqfw5mgawgggbvbc693h833znzrecallvk9784jjqfw5mgawgggbvbc693h833znz

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Runtime requirements

🧠 Clawdis
Any bindocker, uv

Comments