claws-dream
v2.2.2⚠️ DEPRECATED — OpenClaw 2026.4.5+ has official Dreaming (memory-core) built-in. This skill is no longer maintained. Use the official /dreaming slash command...
⭐ 0· 84·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
Name/description (memory consolidation) match the included scripts and documentation. The scripts and Python module operate on the workspace MEMORY.md, memory/ directory, and index.json as expected for a consolidation/indexing tool. No unrelated binaries or credentials are requested.
Instruction Scope
Runtime instructions explicitly read and modify files in the workspace (MEMORY.md, memory/*.md, memory/index.json), mark processed logs, create backups, and generate a dashboard — all within the stated scope. The SKILL.md also instructs the user to change agent timeout in openclaw.json and to create system schedulers (launchd/systemd/cron) which require user/system changes; this is operationally relevant but broadens the surface area and should be applied deliberately. The script triggers an 'openclaw cron run <UUID>' command — plausible for invoking platform run logic but opaque; review that UUID/command if you want to be certain what it triggers.
Install Mechanism
There is no network install step or remote download; the skill is instruction-only with included local scripts and a Python module. Nothing in the install spec writes arbitrary remote code or pulls from unknown URLs.
Credentials
The skill requires no environment variables or external credentials. The scripts optionally use OPENCLAW_WORKSPACE (falling back to a sane ~/.openclaw/workspace); this is appropriate for a local workspace tool and not excessive.
Persistence & Privilege
The skill does not request always:true and is user-invocable only. It instructs users to add scheduled tasks to run the script nightly (system-level cron/systemd/launchd entries) and to increase the agent timeout in openclaw.json — both are legitimate for a long-running nightly job but grant persistent, automated execution on the host if the user configures them. The agent-invocation command ('openclaw cron run <UUID>') is able to trigger platform behavior; confirm the UUID's intent before enabling automation.
Assessment
This skill is coherent with its stated purpose: it reads your workspace memory files and updates MEMORY.md and memory/index.json, makes backups, and can be scheduled to run nightly. Before installing or enabling scheduled runs: 1) Review scripts/dream.sh and scripts/update-index.py to ensure you are comfortable with the exact file paths they read/write; they operate on ~/.openclaw/workspace by default. 2) The skill asks you to change openclaw.json timeout and to create system cron/systemd/launchd tasks — these are manual host changes that grant the skill persistent automated runs; only apply them if you trust the code. 3) Inspect the 'openclaw cron run <UUID>' invocation (in dream.sh) to confirm what platform action that UUID triggers. 4) Because the skill will read private interaction logs and write memory files, consider running it in a test workspace first and keep a copy of MEMORY.md backups before enabling automation. If you prefer an officially supported alternative, the SKILL.md itself says newer OpenClaw versions include built-in dreaming — consider using that instead.Like a lobster shell, security has layers — review code before you run it.
automationvk9786mgkw5nzczk6hndryne3b9840vvhlatestvk97be35a13wkv1nyw0s949m7f984adr6memoryvk9786mgkw5nzczk6hndryne3b9840vvhproductivityvk9786mgkw5nzczk6hndryne3b9840vvh
License
MIT-0
Free to use, modify, and redistribute. No attribution required.