Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/kimsanguine/hplannpx agentmods add commands/kimsanguine/hplan/harness-planWrote 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/commands/kimsanguine/hplan/harness-plan)<a href="https://agentmods.dev/commands/kimsanguine/hplan/harness-plan"><img src="https://agentmods.dev/badge/commands/kimsanguine/hplan/harness-plan.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 5.1 | $0.00049 | $0.04288 |
| Opus 5 | $0.00024 | $0.02144 |
| Sonnet 5 | $0.00010 | $0.00858 |
| Haiku 4.5 | $0.00005 | $0.00429 |
Grade A, and why
harness-plan 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 7d 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.
How it starts
The opening of the file, as written. The whole thing — 352 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/harness-plan
에이전트 시스템 기술 아키텍처 전체 설계 + 전략 리뷰
Routing
$ARGUMENTS에서 --mode 플래그를 파싱한다.
| 플래그 | 실행 범위 |
|---|---|
--mode orchestration |
Phase 1만 실행 |
--mode 3-tier |
Phase 2만 실행 |
--mode memory |
Phase 3만 실행 |
--mode routing |
Phase 4만 실행 |
--mode review |
Phase 5(Strategy Review)만 실행 |
| 플래그 없음 | Phase 1 → 2 → Checkpoint → 3 → 4 → Architecture Doc 전체 실행 (Phase 5는 별도 호출) |
Instructions
You are designing Technical Architecture and Strategy Plan for: $ARGUMENTS
플래그가 있으면 해당 Phase만 실행한다. 플래그가 없으면 아래 전체 플로우를 순서대로 실행한다.
Planning Disciplines (v0.9.6)
Phase 진입 전 아래 3개 규율을 명시적으로 적용한다.
G2 — Named Artifacts: Phase 시작 직후, 이 Phase의 산출물 파일명·섹션명을 먼저 선언한다.
→ 파일명 없이 "설계를 진행하겠습니다"라고 시작하면 규율 위반.
G3 — Decision Commit: 모든 HITL 결정 지점에서 옵션 3개 이상 제시 → 정확히 1개에 커밋.
→ "A 또는 B 방향으로 갈 수 있습니다" 같은 미결 처리 금지. decision_log.py hitl 기록 필수.
G4 — Phase Context Budget: 각 Phase는 해당 Phase에 필요한 파일만 로드한다 (최대 3개).
→ Phase 시작 전 전체 프로젝트 파일 일괄 Read 금지. 필요한 것만 Just-in-Time 로드.
| Phase | Named Artifact | Context Budget (최대) |
|---|---|---|
| 1 Orchestration | decisions/orchestration-choice.md 또는 decision_log 항목 |
없음 (신규 설계) |
| 2 3-Tier | harness/ARCHITECTURE.md — Tier 섹션 |
Phase 1 결정 1개 |
| 3 Memory | harness/ARCHITECTURE.md — Memory 섹션 |
Tier 설계 1개 |
| 4 Routing | harness/ARCHITECTURE.md — Routing 섹션 |
Memory 설계 1개 |
| Architecture Doc | harness/ARCHITECTURE.md 완성본 |
Phase 1–4 결정 요약 |
Phase 1 — Orchestration Pattern Selection
--mode orchestration 또는 전체 플로우
왜: 에이전트 여러 개를 어떻게 연결할지는 나중에 바꾸기 어렵습니다. 시작 전에 Sequential·Parallel·Router·Hierarchical 중 어떤 패턴이 맞는지 결정합니다.
4가지 오케스트레이션 패턴을 사용 사례에 대입해 평가한다.
Orchestration (오케스트레이션): 여러 에이전트가 어떤 순서와 방식으로 협력할지 설계하는 것입니다. 사람 조직의 업무 프로세스 설계와 같은 개념입니다.
| Pattern | Best When | Drawback |
|---|---|---|
| Sequential | 각 스텝이 이전 결과에 의존하고 순서가 고정된 경우 | 느림; 단일 장애 지점 |
| Parallel | 독립적인 서브태스크를 동시에 실행할 수 있는 경우 | 결과 집계가 복잡해짐 |
| Router | 입력 유형에 따라 서로 다른 처리 경로가 필요한 경우 | 라우터 자체가 병목이 됨 |
| Hierarchical | 복잡한 태스크에 계획 레이어와 실행 레이어 분리가 필요한 경우 | 복잡도·비용 최고 |
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.
- 7d ago First seen · 352 lines · 49 tokens per session scan A c33b7dbd6172
harness-plan is a command published in the GitHub repository kimsanguine/hplan (2 stars, last pushed 21d ago), licensed MIT. It adds 49 tokens to every session and 4,288 once invoked, about $0.0002 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 commands, from other repositories
openehr-explain
One-stop router that explains or looks up any openEHR thing — auto-detects an archetype, a template, an RM/AM/BASE type, an RM structural concept, an ADL idiom, an AQL query or keyword, or a terminology code (replaces /archetype-explain, /template-explain, /type-spec, /rm-structure, /adl-idiom, /terminology).
co-change
List files that historically co-change with a target (git log heuristic; COCHANGE-style).
start-design-plan
Start collaborative design process with brainstorming and planning.
start-implementation-plan
Create implementation plan from design document.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.