📦

Audit History

altllm-portal-billing - 2 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v2 LatestJul 21, 2026, 02:34 AM 1 confirmed0No capability change
v1 Jul 9, 2026, 11:49 PM No confirmed findings2Baseline

Jul 21, 2026, 02:34 AM

The documented Portal URL and Markdown command references do not independently indicate unsafe network access or shell execution. However, the skill requires two unscanned shared setup documents; publication should include a review of those dependencies before users follow their instructions.

2
Files scanned
125
Lines analyzed
4
Review items
0
False positives ignored

Confirmed security concerns (1)

Medium
Unreviewed Shared Instruction Dependency
The skill requires reading two shared documents outside its package before running commands. Those documents were not among the audited files, so their instructions cannot be evaluated in this report.
The mandatory dependency is explicit, while only SKILL.md and the CLI reference were scanned. External instructions can affect command execution and should be reviewed with this skill.
Audited by: claude

Jul 9, 2026, 11:49 PM

The hardcoded URL findings are benign documentation for the intended AltLLM Portal API endpoint. The external command findings are false positives from Markdown inline code, but the sibling `../_shared` references are confirmed boundary-crossing filesystem risks. No prompt injection attempt or hidden exfiltration intent was found in the reviewed files.

2
Files scanned
125
Lines analyzed
5
Review items
0
False positives ignored
Capability review items (2)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High
Path traversal sequence
> - `../_shared/preflight.md`
The skill instructs the agent to read a sibling path outside the skill directory. That can import unscanned instructions into the skill context.
High
Path traversal sequence
> - `../_shared/session-and-target.md`
The skill instructs the agent to read another sibling path outside the skill directory. This crosses the marketplace package boundary and should fail closed.
Audited by: codex