openclaw
thedotmack/claude-mem / openclaw
guides installing claude-mem on an OpenClaw gateway , which improves: persistent memory across agent sessions
Details
- judge’s score
- 6/10
- place
- outside the packs
- publisher
- thedotmack
- status
- new
- pinned commit
- 9e4f313 on GitHub (new tab)
- skill’s last commit
- Oct 5, 2026
- repo stars
- 101,903
- main type
- Building with Claude
Judgement
Installation guide for the OpenClaw gateway, not for Claude Code. No frontmatter or description of when to use it. Concrete config and commands, but mostly product documentation.
reservations
- No frontmatter or description: the triggering is not explained.
- Targets OpenClaw, not Claude Code: limited usefulness for Amdy.
- Annexes skills/do and skills/make-plan point to absent files; TESTING.md truncated.
- Duplicates and wording inconsistencies (sentence repeated about the observation, redundant verification steps).
- Triggering
- 1/2
- Added value
- 1/2
- Precision
- 2/2
- Scope
- 1/2
- Quality
- 1/2
Alerts
6 alerts, by category
Static scan
- high risk
downloads and runs a script (curl | sh) (in a code example, to be verified)
- medium risk
symbolic link
skills/do/SKILL.md
- info risk
bundled scripts, not run by the SKILL.md
e2e-verify.sh, install.sh, skills/agent-cost-report/scripts/acr.py, skills/agent-cost-report/scripts/acr/__init__.py, skills/agent-cost-report/scripts/acr/behavior.py
Network
- high risk
Observations (outputs of Read, Bash tools, etc.) are sent to the worker then to a third-party messaging channel (Telegram, Discord, Slack...) with no secret filtering mentioned: possible leak of code or credentials.
- high risk
Session contents are sent to an AI provider chosen at installation, with no detail on what leaves the machine.
Data
- medium risk
The example with --api-key=YOUR_KEY on the command line exposes the key in the shell history and the process list.
Original description
—
added on Oct 19, 2026 · license Apache-2.0