aeondave/vec2text
ai/vec2text/SKILL.md
vec2text: embedding-inversion library for reconstructing approximate text from sentence embeddings. Use when working with saved embedding tensors, privacy/inversion research, or AI/ML challenge workflows where you need to load a corrector model and invert strings or embeddings directly.
$ install --global
skillsauthnpx skillsauth add aeondave/malskill vec2textInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 8, 2026, 3:46 AM138.8s1 file scanned
SKILL.md
- name:
- vec2text
- description:
- vec2text: embedding-inversion library for reconstructing approximate text from sentence embeddings. Use when working with saved embedding tensors, privacy/inversion research, or AI/ML challenge workflows where you need to load a corrector model and invert strings or embeddings directly.
- compatibility:
- Linux, macOS, Windows; Python 3; PyTorch and Hugging Face stack; GPU strongly recommended for larger runs
- author:
- AeonDave
- version:
- 1.0
vec2text
Use vec2text when embeddings are the artifact and text recovery is the question.
When to use vec2text
Use vec2text when you need to:
- invert sentence embeddings back into approximate text
- load a pretrained corrector for supported embedding families
- work from saved embedding tensors rather than original source text
- tune inversion quality with multiple refinement steps or beam search
Quick Start
import vec2text
corrector = vec2text.load_pretrained_corrector("gtr-base")
High-Value Workflows
Invert known strings through the matching embedder workflow
vec2text.invert_strings(
["example text"],
corrector=corrector,
num_steps=20,
sequence_beam_width=4,
)
Invert embeddings directly
import torch
embeddings = torch.load("embeddings.pt", map_location="cpu")
texts = vec2text.invert_embeddings(
embeddings=embeddings,
corrector=corrector,
num_steps=20,
)
print(texts)
Practical Notes
- Corrector choice must match the embedding family; mismatched models produce garbage.
num_stepsimproves refinement quality, whilesequence_beam_widthimproves search at higher memory cost.- The project supports both direct string inversion workflows and embedding-only inversion workflows.
Caveats
- Reconstruction is approximate, not guaranteed exact text recovery.
- Beam search raises memory usage quickly.
- Model support and pretrained aliases evolve; prefer current upstream docs over old examples.
Resources
No bundled scripts/, references/, or assets/.
Use the upstream vec2text README for current pretrained corrector aliases, API entry points, and embedding-family matching guidance.
Related Skills
aeondave/offensive-linux-role
data-ai
VerifiedTrustedCommunity
Scoped routing: Linux operator; hosts, sessions, users, services, packages, logs, containers, SSH, network paths, privilege evidence.
4SKILL.mdUpdated Jun 5, 2026
aeondave/ics-technique
development
VerifiedTrustedCommunity
Offensive methodology for ICS/OT/SCADA environments in authorized industrial penetration testing and red team operations. Use when assessing PLCs, RTUs, HMIs, engineering workstations, historians, or field devices running Modbus, DNP3, EtherNet/IP, S7comm/S7+, Profinet, IEC 60870-5-104, BACnet, or OPC-UA. Covers passive OT network enumeration, protocol-level device interrogation, PLC coil/register read-write attacks, HMI session exploitation, historian and engineering workstation compromise, and safe escalation rules for critical infrastructure scope. Does not cover: general IT network exploitation (network-technique), physical hardware interfaces UART/JTAG/SPI (hardware-technique), wireless sensor network attacks (wireless-technique), RF/SDR signal analysis (hardware-ctf or wireless-technique), or CTF-framed ICS lab tasks (ics-ctf).
4SKILL.mdUpdated Jun 2, 2026
aeondave/game-technique
tools
VerifiedTrustedCommunity
Offensive methodology for authorized game security assessments, game client security research, and game-adjacent penetration testing in real-world engagements. Use when assessing game clients for cheating vulnerabilities, testing anti-cheat effectiveness, auditing game server protocols for score manipulation or economic fraud, reverse engineering game DRM or license validation, analyzing game save file protection, or assessing game mod/plugin security. Covers: process memory scanning and manipulation (Cheat Engine methodology), game binary reversing for license and DRM bypass, game network protocol analysis and packet replay, anti-cheat mechanism analysis, save file format reversing and tampering, speed hack and value injection techniques. Does NOT cover: CTF game challenges (game-ctf), game engine source code auditing (web-exploit-technique or vuln-search-technique for the backend), or general binary exploitation (pwn-ctf or reversing-technique).
4SKILL.mdUpdated Jun 2, 2026
aeondave/hardware-technique
development
VerifiedTrustedCommunity
Auth assessment: hardware/embedded methodology; UART/JTAG/SWD/SPI/I2C, firmware extraction, boot/debug paths, embedded OS evidence.
4SKILL.mdUpdated Jun 2, 2026