governed-autonomy-loop

Warn

Audited by Socket on Jul 11, 2026

2 alerts found:

AnomalySecurity
AnomalyLOW
scripts/tick.sh

No direct evidence of classic supply-chain malware (e.g., hidden exfiltration, reverse shells, cryptomining, hardcoded secrets) is present in this Bash fragment. However, there are significant security risks: (1) config.env is shell-sourced ('. "$CONFIG"'), which can allow arbitrary command execution if config.env is modified; and (2) the script spawns a configurable external binary (CLAUDE_BIN) with '--dangerously-skip-permissions' and a dynamically constructed prompt that includes many operational parameters and paths. These behaviors make the overall system highly sensitive to tampering of config.env, DENYLIST_FILE, or GAL_CLAUDE_BIN/REPO_DIR content. Overall, the static snippet appears more like an orchestration/scheduling component than a malicious payload, but it has high blast-radius if an attacker can influence its inputs.

Confidence: 68%Severity: 55%
SecurityMEDIUM
SKILL.md

SUSPICIOUS: the skill is internally coherent, but it enables unattended autonomous operations with real-world consequences and broad operational authority. No clear malicious install path or hidden exfiltration is shown, yet the autonomy scope and credentialed adapter surface make it high security risk.

Confidence: 84%Severity: 78%
Audit Metadata
Analyzed At
Jul 11, 2026, 04:28 AM
Package URL
pkg:socket/skills-sh/broomva%2Fskills%2Fgoverned-autonomy-loop%2F@5943404bdd64b05ad283b25ec0ff82e485d9fb4463d3a38a7c1862adffb8f451
Security Audit — socket — governed-autonomy-loop