🟡 [MEDIUM] CB-012: No Rate Limiting on Tool Execution or LLM Calls #21

New issue

Open

opened 2026-02-20 08:50:04 +00:00 by Alpha · 0 comments

Alpha commented

2026-02-20 08:50:04 +00:00

Security Finding from Audit

Parent issue: #1

Severity

MEDIUM 🟡

CVSS Score

N/A

CWE

CWE-770 (Allocation of Resources Without Limits)

Location

agent.py, respond(), tool execution loop

Description

The agent allows up to 10 tool calls per message with no global rate limiting.

An attacker can trigger expensive LLM calls (draining API credits) or rapid tool execution (fork bomb via exec, disk filling via write_file) through repeated messages.

Recommendation

Implement per-user, per-channel, and global rate limits. Add a cost ceiling per interaction.

From: Cobot Whitebox Security Audit (February 14, 2026)
Finding ID: CB-012

## Security Finding from Audit > Parent issue: #1 ### Severity **MEDIUM** 🟡 ### CVSS Score N/A ### CWE CWE-770 (Allocation of Resources Without Limits) ### Location `agent.py, respond(), tool execution loop` --- ### Description The agent allows up to 10 tool calls per message with no global rate limiting. An attacker can trigger expensive LLM calls (draining API credits) or rapid tool execution (fork bomb via exec, disk filling via write_file) through repeated messages. --- ### Recommendation Implement per-user, per-channel, and global rate limits. Add a cost ceiling per interaction. --- *From: Cobot Whitebox Security Audit (February 14, 2026)* *Finding ID: CB-012*