n2-fieldops
n2-fieldops is a dual-runtime Agent Skills distribution for Claude Code, Codex GPT,
and CC Switch. It combines a hardened runtime profile, 17 self-contained skills,
deterministic helper scripts, and an evidence-first CTF layer adapted from the pinned
ljagiello/ctf-skills snapshot.
The repository has one canonical skill tree. Every installable skill is a root-level
directory containing SKILL.md. This is deliberate: current CC Switch recursively
scans repositories, so a second nested copy would create duplicate cards and
unpredictable install collisions.
Install directly with CC Switch
In current CC Switch releases, open Skills -> Manage Skill Repositories -> Add Repository and enter:
| Field | Value |
|---|---|
| Repository URL | https://github.com/download4you/n2-fieldops |
| Branch | main |
CC Switch also accepts the compact repository form:
download4you/n2-fieldops
Refresh the catalog and install the skills for the Claude and/or Codex application
tabs. No subdirectory is required or supported by the current repository form. The
repository intentionally exposes exactly 17 root skills. If an older
ljagiello/ctf-skills or ZIP-based copy is enabled, disable or remove it first to
avoid duplicate routing and stale content.
CC Switch installs the same skill directories into the selected app's skill store;
the runtime profile remains a separate, reviewed file. For Codex, copy AGENTS.md
to the project root or to $HOME/.codex/AGENTS.md. For Claude, review and copy
claude-plugin-template/docs/CLAUDE_PROFILE.md as CLAUDE.md when an always-on
profile is desired.
Install with the Agent Skills CLI
npx skills add download4you/n2-fieldops
Claude native plugin release
The native Claude plugin is generated from the same canonical tree during release and
published as n2-fieldops-<version>-claude-plugin.zip. The marketplace bundle is
published as n2-fieldops-<version>-claude-marketplace.zip. This keeps the GitHub
repository safe for CC Switch's recursive scanner while still providing a standard
Claude plugin for local development or a private marketplace.
Build the artifacts locally:
py -3 scripts/build_release.py --output-dir dist
claude --plugin-dir .\dist\n2-fieldops-claude-plugin
The 17 skills
| Skill | Purpose |
|---|---|
fieldops-orchestrator |
Route complex or mixed-domain work across the smallest useful skill set |
fieldops-research |
Verify uncertain or current claims and deliver evidence-backed answers |
fieldops-prompt-decorators |
Parse and apply stateful +++ controls without executing untrusted tokens |
fieldops-ctf-operator |
Triage, route, and reproduce authorized CTF investigations |
fieldops-engineering |
Diagnose, implement, and verify consequential repository changes |
fieldops-powershell-utf8 |
Preserve Persian and other Unicode across PowerShell boundaries |
fieldops-prompt-refiner |
Turn rough prompts and agent instructions into execution-ready briefs |
fieldops-ctf-ai-ml |
Analyze adversarial ML, model extraction, and LLM challenge surfaces |
fieldops-ctf-crypto |
Attack classical and modern cryptographic constructions |
fieldops-ctf-forensics |
Investigate disk, memory, packet, signal, stego, and artifact evidence |
fieldops-ctf-malware |
Analyze malware, C2 traffic, obfuscation, and anti-analysis behavior |
fieldops-ctf-misc |
Solve encoding, jail, RF, Unicode, game, and hybrid puzzles |
fieldops-ctf-osint |
Perform competition-scoped public-source and geolocation research |
fieldops-ctf-pwn |
Develop native, heap, kernel, sandbox, and exploit-chain solutions |
fieldops-ctf-reverse |
Reverse binaries, firmware, bytecode, VMs, and packed targets |
fieldops-ctf-web |
Analyze web applications, APIs, browser state, and identity flows |
fieldops-ctf-writeup |
Produce reproducible submission-style challenge documentation |
For a difficult or mixed task, start with fieldops-orchestrator. For an authorized
CTF whose category is unclear, start with fieldops-ctf-operator; it routes to the
specialist and requires clean-baseline reproduction before claiming a solve.
Validation and release build
py -3 -m unittest discover -s tests -v
py -3 scripts/build_release.py --check
py -3 scripts/build_release.py --output-dir dist
The checks cover frontmatter, CC Switch discovery, self-contained references, provenance, decorator parsing, router determinism, plugin generation, and the PowerShell 5.1/7 UTF-8 round trip.
Model guidance
The prompts are model-neutral and are written for the current Claude Opus family and
current Codex GPT/Codex models. Select the model exposed by the client; no stale
version-specific model identifier is required. The important portability boundary is
the adapter: AGENTS.md is the Codex profile, while the reviewed Claude profile is
claude-plugin-template/docs/CLAUDE_PROFILE.md.
Provenance and hardening
The CTF specialist material is adapted from
ljagiello/ctf-skills at pinned commit
d19f35fd3dd2e126108752aee84c657c888126d3 under MIT. Prompt Decorators are adapted
from smkalami/prompt-decorators under
MIT. See THIRD_PARTY_NOTICES.md and
docs/PROVENANCE.md.
The local Codex .system skills (for example image generation, OpenAI docs,
plugin/skill creators, and review helpers) remain runtime-owned capabilities and are
not copied or redistributed. Their useful integration boundaries are documented in
docs/CODEX_INTEGRATION.md.
The source prompt transcripts were used as local design inputs, not as third-party code. Unsafe blanket-authorization and fictional-network clauses were intentionally removed; the shipped profiles require clear authorization for security work and do not promise hidden chain-of-thought disclosure.
Version lineage
The n2-fieldops product line starts at v1.0.0; v1.0.1 is the reproducibility and
installation-documentation maintenance release. Older v2.x and v3.0.0 releases
visible in this repository belong to the predecessor Codex FieldOps project and are
retained as provenance history. Use the main branch or GitHub's latest n2-fieldops
release rather than selecting the numerically highest legacy tag.
License
Original n2-fieldops material is released under MIT. Adapted third-party material remains subject to its upstream notices.
No comments yet
Be the first to share your take.