SilicaClaw Bridge Setup
Analysis
This is a coherent local setup guide for SilicaClaw, but it asks you to run installer and forwarding commands that can change OpenClaw’s installed skills and notification destination.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Install the bundled skills into OpenClaw: `silicaclaw openclaw-skill-install` ... installed under `~/.openclaw/workspace/skills/`
This is a purpose-aligned setup command, but running it can add persistent skills to the OpenClaw workspace and change future agent behavior.
Source: unknown; Homepage: none
The reviewed registry metadata does not provide an upstream source or homepage, so the provenance of the local SilicaClaw tooling should be checked before following installer commands.
Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.
export OPENCLAW_OWNER_CHANNEL="<channel>" ... export OPENCLAW_OWNER_TARGET="<target>" ... export OPENCLAW_OWNER_FORWARD_CMD="node scripts/send-to-owner-via-openclaw.mjs"
These variables control private owner delivery and the local command used to forward messages, which is expected for owner notifications but should be tightly owner-approved.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
"api_base_env": "SILICACLAW_API_BASE", "default_api_base": "http://localhost:4310"
The bridge uses a local HTTP endpoint by default and can be redirected through an environment variable, which is coherent for a local bridge but still defines a data boundary.
