Openclaw Agent Optimize
v1.2.1Use when: you want a structured audit -> options -> recommended plan to improve an OpenClaw workspace (cost, model routing, context discipline, delegation, r...
⭐ 47· 10.6k·107 current·108 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (agent optimization: cost, routing, context discipline) matches the contents: SKILL.md + references provide guidance for audits, cron/heartbeat changes, model-tiering, and rollback plans. The skill requests no credentials, binaries, or config paths, which is proportional for an advisory auditor.
Instruction Scope
Runtime instructions are advisory-first: perform audits, propose exact patches, and require explicit user approval before applying persistent changes (crons, agent defaults). The docs reference reading/writing workspace artifacts (memory/, references/, logs) and proposing config patches — these are expected for this purpose and are explicitly gated by an approval requirement. There are no instructions to exfiltrate secrets or to contact external endpoints.
Install Mechanism
This is instruction-only (no install spec, no code files executed). README suggests installing via a third-party ClawHub tool or copying the folder manually; those are optional distribution channels and not part of an automated install that would fetch arbitrary binaries. No archive downloads or third-party package installs are embedded in the skill itself.
Credentials
The skill declares no required environment variables, no primary credential, and no config paths. References talk about local workspace files (memory/, references/, logs) which are appropriate for an optimization/audit skill. A recommendation to consider installing an optional helper (openclaw-mem) is informational and not a demanded credential.
Persistence & Privilege
Skill metadata: always=false and user-invocable=true. SKILL.md repeatedly instructs not to mutate persistent settings or create/edit cron jobs without explicit approval and to include exact diffs, rollback, and verification steps. Autonomous invocation (model invocation allowed) is the platform default and is not combined with any unusual privileges here.
Assessment
This skill is advisory and appears internally consistent: it will read workspace artifacts (memory/, references/, logs) and produce proposed config patches and cron changes but explicitly forbids applying persistent changes without your approval. Before approving any change: review the exact diffs, confirm the rollback steps, verify file/write locations, and confirm you trust the source (ClawHub page or repository). Do not provide external secrets or grant other credentials; treat any recommendation to install additional helpers (e.g., openclaw-mem) as optional and review that project's code separately.Like a lobster shell, security has layers — review code before you run it.
latestvk979qy4y26s9j4ab497tzp285n82hdqd
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧰 Clawdis