Caveat verdict
sparkbtcbot
Fetches from network AND uses eval/Function
The skill explicitly sets model-invocation to autonomous and states agents can send funds 'without human approval for each transaction' using a mnemonic that controls all wallet funds — while the design is transparent and recommends the proxy for production, the default configuration grants the agent unilateral control over real Bitcoin.
Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
What it does
These are capability combinations: each listed behavior occurs in the skill, but Caveat detects co-occurrence — it does not verify that one flows into another. Read the code to confirm a live chain.
Fetches from network AND uses eval/Function — the remote-code-execution pattern (data-flow not verified)
LLM05 · LLM06 · ASI05
Receives external input AND uses eval — the remote code-injection pattern (data-flow not verified)
LLM01 · LLM05 · ASI01 · ASI05
Encodes data AND uses eval — the obfuscated-execution pattern (atob + eval; data-flow not verified)
LLM05 · ASI05 · ASI10
Accesses credentials AND makes external network calls — potential credential theft
LLM02 · ASI03
Accesses credentials AND encodes data — may obfuscate stolen credentials
LLM02 · ASI03 · ASI04
Permission integrity
network_out
credential_access
package_install
Findings (5)
Dynamic import() — loads modules at runtime
SKILL.md · code
Dynamic import() — loads module at runtime
SKILL.md · code · import("
Instructs covert action — may act without user awareness
BUGFIX-tokenBalances-map.md · code · silently
Accesses process.env — reads environment variables
SKILL.md · code
fetch() — outbound network request
SKILL.md · code
Permissions & capabilities
No declared permissions — minimal attack surface.
package_installcredential_accessdata_encodingnetwork_outnetwork_indynamic_eval Thanks — recorded.