Skills new-term-radar Audit History
📦

Audit History

new-term-radar - 3 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v3 LatestOct 28, 2026, 12:29 PM No confirmed findings0No capability change
v2 Oct 27, 2026, 09:12 AM 1 confirmed0No capability change
v1 Oct 20, 2026, 02:46 PM 1 confirmed0Baseline

Oct 28, 2026, 12:29 PM

All 50 static findings are false positives involving Markdown formatting, ordinary authentication requirements, public reference links, and unrelated prose. The instructions require paid-run approval and treat external content as untrusted; no evidence found of credential theft, destructive intent, or prompt injection. This review covers SKILL.md only, not the external PyPI package or its dependencies.

1
Files scanned
93
Lines analyzed
4
Review items
0
False positives ignored
Audited by: codex

Oct 27, 2026, 09:12 AM

Static matches are Markdown, disclosed service references, or research terminology, not evidence of the alleged attacks. The external package runs with service credentials, but its implementation was not reviewed.

1
Files scanned
89
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (1)

Medium
Unaudited external package executes with service credentials
The instructions fetch and execute a PyPI package with access to service credentials. The package implementation is outside the reviewed skill file.
The document explicitly invokes the package and requires a credential. Its implementation and dependency integrity were not reviewed.
Audited by: codex

Oct 20, 2026, 02:46 PM

All 42 static alerts are false positives: Markdown formatting, documented service authentication, public links, and research terminology explain the matches. No evidence found of credential theft, host reconnaissance, or prompt injection in the supplied instructions. A separate medium-risk concern remains: setup clones an unpinned external repository and executes its unaudited code with access to configured credentials.

1
Files scanned
77
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (1)

Medium
Unpinned External Code Execution
Setup says to clone https://github.com/zhuyansen/new-term-radar if missing, then run Python modules from that directory. No reviewed revision or integrity check is specified. A changed upstream checkout could execute unreviewed code with configured credentials.
The instructions explicitly combine an unpinned repository clone, credential availability, and local Python execution. Only SKILL.md was supplied; runtime safety cannot be established.
Audited by: codex