Skills ai-listening Audit History
📦

Audit History

ai-listening - 3 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v3 LatestOct 28, 2026, 12:25 PM No confirmed findings0No capability change
v2 Oct 27, 2026, 09:11 AM No confirmed findings4No capability change
v1 Oct 20, 2026, 02:43 PM No confirmed findings2Baseline

Oct 28, 2026, 12:25 PM

All 24 static findings are false positives involving Markdown formatting, documented CLI usage, expected authentication checks, reference links, or report filtering. No evidence found of credential exfiltration, system reconnaissance, or prompt injection; the source explicitly rejects instructions embedded in collected data. This review covers SKILL.md only, not the external package implementation or its runtime network behavior.

1
Files scanned
60
Lines analyzed
4
Review items
0
False positives ignored
Audited by: codex

Oct 27, 2026, 09:11 AM

The skill instructs users to execute a pinned third-party package and optionally provide a service credential. The package implementation was not included in the reviewed file, so its behavior and credential handling remain unverified. Other flagged backticks, links, credential-path checks, and reconnaissance patterns are benign in context.

1
Files scanned
56
Lines analyzed
8
Review items
0
False positives ignored
Capability review items (4)

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
Generic API/secret keys
3. Check an AIsa key is available: `AISA_API_KEY`, or the AIsa CLI logged in (`ls ~/.aisa/key`). Wit
The workflow uses an API key or local login to run an external package. The package implementation was not reviewed, so credential handling cannot be verified.
Medium
Ruby/shell backtick execution
1. Run the pinned release, never a clone of the repository: `uvx new-term-radar==0.6.0 <command> ...
The instructions execute a PyPI package through uvx or pip. Version pinning does not verify package integrity, and the package code was not reviewed.
Medium
Ruby/shell backtick execution
```bash
The fenced block instructs execution of a downloaded PyPI package. Its implementation is outside the reviewed skill file.
Medium
Ruby/shell backtick execution
If the user only wants the fastest-moving sources, `uvx new-term-radar==0.6.0 sources --out listenin
The optional sources command runs the same downloaded PyPI package, whose code is outside this review.
Audited by: codex

Oct 20, 2026, 02:43 PM

Most alerts match Markdown formatting, documented authentication checks, or report fields rather than dangerous behavior. The fallback setup clones an unpinned external repository and subsequently executes its Python module, creating a supply-chain risk. No evidence found of credential exfiltration or prompt injection in SKILL.md; the external implementation was not included for review.

1
Files scanned
51
Lines analyzed
6
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.

Medium
Ruby/shell backtick execution
1. Find the radar. Use `$NEW_TERM_RADAR_DIR` if set, else `~/new-term-radar`. If neither exists, clo
Setup clones a mutable repository without a pinned revision, then directs commands to run there. The subsequent Python invocation executes unaudited external code.
Low
Hardcoded URL
1. Find the radar. Use `$NEW_TERM_RADAR_DIR` if set, else `~/new-term-radar`. If neither exists, clo
The GitHub URL supplies executable code through an unpinned clone, followed by local execution. The risk is dependency integrity, not demonstrated maliciousness of GitHub.
Audited by: codex