Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Kannaka Eye
v1.0.1Glyph viewer that renders the SGA geometric fingerprint of any data as a stunning multi-layer canvas visualization. Takes text, files, or raw bytes and produ...
⭐ 0· 308·0 current·0 all-time
byNick Flach@nickflach
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The SKILL.md claims a single-file Node.js server (server.js) that provides upload/processing endpoints, but the package does not include server.js or any JS server source — only a CLI wrapper (scripts/eye.sh), README, SKILL.md and metadata. _meta.json lists a dependency on 'flux' and optional binaries while the actual code implementing network behavior is absent. This mismatch suggests the published bundle is incomplete or altered; you cannot verify the server's behavior from the provided files.
Instruction Scope
The runtime instructions tell the agent/user to run ./scripts/eye.sh which invokes node server.js; since server.js is not present we cannot validate the described endpoints (/api/process, /api/radio, /api/constellation) or confirm claims such as "uploads processed in-memory only". The docs also describe optional publishing of glyph.rendered events to a Flux URL (FLUX_URL) — enabling that would transmit data off-host. The SKILL.md references auto-detection of a native binary at ../kannaka-memory/target/release/kannaka.exe which reaches outside the repo and could cause the service to invoke external binaries if present.
Install Mechanism
No install spec (instruction-only) and a small shell wrapper are provided, which is low-risk in itself. However, because the main server code is missing from the bundle, the claimed "zero dependencies" server cannot be validated; if the real server is fetched or replaced at runtime, that would increase risk. The provided scripts themselves are benign shell wrappers that start/stop node, use lsof/netstat to find a process, and call kill.
Credentials
Declared requirements are minimal (node only) and optional environment variables (KANNAKA_BIN, EYE_PORT, RADIO_PORT, FLUX_URL, FLUX_AGENT_ID) align with the described functionality. The primary proportionality concern is FLUX_URL: if set, the skill would (per its docs) publish events remotely, which could leak processed or uploaded data. The auto-detect path for KANNAKA_BIN points outside the project and could cause the service to execute a local native binary if present.
Persistence & Privilege
The skill is not always-enabled and is user-invocable (defaults). There is no install step that writes persistent system-wide config in the provided files. Nothing in the bundle requests elevated or persistent privileges. Autonomous invocation is allowed by default but not accompanied here by other high-risk signals.
What to consider before installing
Do not run this skill as-is. The package is missing the server.js implementation the README and SKILL.md describe, so you cannot audit network or file-handling behavior. Ask the publisher for the full server source (server.js) and verify it locally before running. If you do run it, keep FLUX_URL unset unless you trust the remote endpoint (setting it would cause the skill to publish glyph events remotely and could leak processed input). Also verify any native KANNAKA_BIN auto-detection path and review what that binary does before allowing it to be used. If you cannot obtain and review server.js, treat the skill as incomplete/untrusted.Like a lobster shell, security has layers — review code before you run it.
latestvk97cjm369zrefppm961g0z58nd82mm5j
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
Bins[object Object]