Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/tellang/triflux/agents-mdgit clone --depth 1 https://github.com/tellang/trifluxWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/tellang/triflux/agents-md)<a href="https://agentmods.dev/instructions/tellang/triflux/agents-md"><img src="https://agentmods.dev/badge/instructions/tellang/triflux/agents-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00827 | $0.00827 |
| Opus 5 | $0.00413 | $0.00413 |
| Sonnet 5 | $0.00165 | $0.00165 |
| Haiku 4.5 | $0.00083 | $0.00083 |
Grade A, and why
triflux AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
triflux — Codex 가이드
상세 운영 지시는 CLAUDE.md에 있습니다. Codex는 @import를 지원하지 않으므로 필요할 때 직접 읽습니다.
핵심 규칙 (triflux 환경 특화)
- Codex 직접 호출 금지 —
headless-guard가 차단한다.tfx-auto --cli codex/tfx-multi/tfx-swarm을 거친다. - 프롬프트는
--뒤의 인자로 넘기고 표준 입력은 닫는다(… -- "$prompt" < /dev/null).codex < prompt.md는 비대화식 실행이라 실패한다. config.toml에approval_mode,sandbox기본값을 두고 CLI는--profile만 지정한다. 중복 플래그를 쓰지 않는다.- Claude가 작성한 코드는 Codex로 교차 검증한다. 스스로 승인하지 않는다.
- 비대화식 결과는 작업 알림이 완료된 뒤에만 읽는다.
비대화식 실행 보호기(headless-guard)
codex exec/agy --dangerously-skip-permissions --print=직접 호출은 차단됩니다.- 반드시 tfx 스킬인
tfx-auto --cli codex,tfx-multi,tfx-swarm을 거친다.
추론 수준 예외 경로
- 기본 역할 라우팅은
low/medium/high/xhigh프로필을 유지합니다. - 최난도 단일 작업만
TFX_CODEX_PROFILE=max(gpt56_sol_max). ultra는 최상위 단독deep-executor/scientist-deep에서만 허용하고, 팀·비대화식·워커 내부의ultra요청은 중첩 분산 실행을 막기 위해gpt56_sol_max로 낮춥니다.--retry auto-escalate의 Codex 단계도gpt56_sol_max입니다.
비대화식 결과 회수
- 완료 마커:
=== HEADLESS_COMPLETE succeeded=N failed=N total=N === - 워커 상세:
$TMPDIR/tfx-headless/{sessionName}-worker-N.txt
SSH 방식
hosts.json의os필드로 셸을 판단합니다 (windows=PowerShell, darwin=zsh, linux=bash).- SSH 너머로 Codex를 직접 실행하지 않는다(
config.toml충돌과 TTY 문제). 원격에서 Codex가 필요하면tfx-remote→ Claude Code → Claude가 내부에서 codex 호출.
작업 합의
- 한국어 우선, 기술 용어는 원어 유지.
- 커밋:
Type: 한국어 설명 (50자 이내).Co-Authored-By와 AI 꼬리표를 넣지 않는다. - 변경 후 린트·타입 검사·테스트를 실행한다. 비밀 정보(API 키·토큰·세션)를 하드코딩하지 않는다.
safety-guard에 새 차단 규칙을 추가할 때는 항상 대안 API를 함께 만든다. 차단만 추가하면 교착 상태가 된다.
CTO 기준 방향
.triflux/lake/current.md는 에이전트 간 기준 방향 요약입니다. 정렬을 위해 읽되 새 작업으로 취급하지 않습니다.- 프로그램용 정본은
tfx cto status --json입니다. 안정 키는schema_version,repo,sources,ledger_tail,live_sessions,active_shards입니다. - 읽기 전용 정렬 문맥입니다.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 42 lines · 827 tokens per session scan A 13f06d9874aa
triflux AGENTS.md is an instructions file published in the GitHub repository tellang/triflux (7 stars, last pushed 14d ago), licensed MIT. It adds 827 tokens to every session, about $0.0041 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
neo4j-skills AGENTS.md
AGENTS.md instructions for neo4j-contrib/neo4j-skills, covering writing skills for this repository, language style — write for agents, not humans, six design principles, skill.md spec (agentskills.io) and frontmatter.
apm python.instructions.md
Python development guidelines.
OpenMicro AGENTS.md
Instructions for stephenleo/OpenMicro, covering openmicro — project conventions and releases.
dsh-codex-app-server AGENTS.md
Instructions for LyleMi/dsh-codex-app-server, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
shift-log CLAUDE.md
Claude Code instructions for re-cinq/shift-log, covering openspec instructions, integration tests and beads task tracking.
Lossless-Codex-Orchestrator-LCO AGENTS.md
AGENTS.md instructions for 100yenadmin/Lossless-Codex-Orchestrator-LCO, covering lco agent instructions, repository agent quick start, shared owned-repo policy, lco release gates and public documentation placement.