Audit History
brewpage-publish - 10 audits
Version comparison
Capability and finding changes across audited versions, newest first.
| Version | Date | Result | Review items | Change vs previous |
|---|---|---|---|---|
| v10 Latest | Nov 19, 2026, 07:38 AM | 4 confirmed | 11 | No capability change |
| v9 | Oct 27, 2026, 05:10 AM | No confirmed findings | 0 | No capability change |
| v8 | Oct 27, 2026, 04:38 AM | 1 confirmed | 0 | No capability change |
| v7 | Oct 27, 2026, 04:04 AM | 1 confirmed | 0 | No capability change |
| v6 | Oct 26, 2026, 10:31 PM | 1 confirmed | 0 | No capability change |
| v5 | Oct 26, 2026, 08:38 PM | 4 confirmed | 0 | Env variables |
| v4 | Jul 5, 2026, 06:06 PM | 4 confirmed | 31 | No capability change |
| v3 | Jul 5, 2026, 06:06 PM | 4 confirmed | 31 | Env variables |
| v2 | Jun 30, 2026, 05:48 AM | 3 confirmed | 0 | Env variables |
| v1 | Jun 2, 2026, 12:56 PM | No confirmed findings | 3 | Baseline |
Nov 19, 2026, 07:38 AM
The reviewed helpers send publication data to BrewPage and store reusable owner credentials locally. No evidence of malicious exfiltration, obfuscation, or prompt injection was found, but approval, archive review, private output, and receipt validation have gaps. The skill was reviewed without executing its scripts or contacting BrewPage.
Confirmed security concerns (4)
Capability review items (11)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
🌐 Network access (25)
🔑 Env variables (9)
⚙️ External commands (50)
📁 Filesystem access (50)
Oct 27, 2026, 05:10 AM
All 272 static alerts reflect documented publishing behavior, defensive checks, or syntax misclassification. No evidence found of unauthorized exfiltration, credential harvesting, hidden execution, or audit-directed prompt injection. Selected content leaves the machine for brewpage.app; unlisted links require passwords for access control.
Risk Factors
🌐 Network access (14)
⚙️ External commands (50)
📁 Filesystem access (37)
Oct 27, 2026, 04:38 AM
All 264 static alerts are false positives involving documented publishing, literal Markdown, validation rules, or local safeguards. A separate directory-packaging race can upload unintended files if a concurrent writer replaces validated paths with symlinks. No evidence found of prompt injection, intentional credential theft, or concealed upload destinations.
Confirmed security concerns (1)
Risk Factors
🌐 Network access (14)
⚙️ External commands (50)
📁 Filesystem access (33)
Oct 27, 2026, 04:04 AM
All 261 static findings are false positives involving documentation, defensive checks, or intended publishing operations. No evidence found of credential theft, encoded payloads, or audit-directed prompt injection. One semantic risk remains: a concurrently modified source ZIP can bypass archive inspection and sensitive-entry confirmation before upload.
Confirmed security concerns (1)
Risk Factors
🌐 Network access (14)
⚙️ External commands (50)
📁 Filesystem access (32)
Oct 26, 2026, 10:31 PM
All 260 static alerts match documentation, exclusion patterns, intended service endpoints, or defensive local operations. A separate high-severity multipart path injection can upload unintended local files through crafted filenames. No evidence found of prompt injection or deliberate credential exfiltration.
Confirmed security concerns (1)
Risk Factors
🌐 Network access (14)
⚙️ External commands (50)
📁 Filesystem access (32)
Oct 26, 2026, 08:38 PM
The static alerts match documentation, defensive exclusions, and expected publishing operations rather than demonstrated malicious behavior. Context review found unsafe history destinations, conflicting token-disclosure instructions, unattended privacy defaults, and unbounded ZIP inspection. No evidence found of deliberate credential theft or encoded executable payloads.
Confirmed security concerns (4)
Risk Factors
🌐 Network access (14)
⚙️ External commands (50)
📁 Filesystem access (28)
🔑 Env variables (4)
Jul 5, 2026, 06:06 PM
Most static hits are Markdown formatting, documentation URLs, or protective exclusions rather than executable risk. Real issues remain in the Bash publish blocks, which interpolate user-controlled values before curl or zip execution and persist owner tokens in a workspace file.
Confirmed security concerns (4)
Capability review items (31)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
🌐 Network access (14)
⚙️ External commands (103)
📁 Filesystem access (4)
Jul 5, 2026, 06:06 PM
Most static hits are Markdown formatting, documentation URLs, or protective exclusions rather than executable risk. Real issues remain in the Bash publish blocks, which interpolate user-controlled values before curl or zip execution and persist owner tokens in a workspace file.
Confirmed security concerns (4)
Capability review items (31)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
🌐 Network access (14)
⚙️ External commands (103)
📁 Filesystem access (4)
Jun 30, 2026, 05:48 AM
Static analysis flagged many shell, network, filesystem, and sensitive-file patterns. Review found these patterns are largely intentional publishing instructions, with no evidence of prompt injection or hidden malicious behavior, but the skill sends user-selected content to an external public hosting service and stores owner tokens locally.
Confirmed security concerns (3)
Static false positives ignored (2)
These static matches were dismissed by semantic review or matched schema-only tokens, so they are shown for transparency but do not drive the quality score.
Risk Factors
🌐 Network access (8)
⚙️ External commands (7)
📁 Filesystem access (5)
🔑 Env variables (4)
Detected Patterns
Jun 2, 2026, 12:56 PM
The skill is a legitimate publishing tool for brewpage.app. All network requests target the documented brewpage.app API endpoints. External commands use standard tools (curl, jq, zip, mktemp) for content upload. The skill implements security safeguards: it excludes .env, .git, node_modules, and other sensitive paths from site uploads, never prints owner tokens in conversation, and stores them in a private file. The static scanner flagged many false positives because the skill's core function inherently requires network access and command execution. No malicious intent detected.
Capability review items (3)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.