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-discoverWrote 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-discover)<a href="https://agentmods.dev/commands/kimsanguine/hplan/harness-discover"><img src="https://agentmods.dev/badge/commands/kimsanguine/hplan/harness-discover/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/kimsanguine/hplan/harness-discover"><img src="https://agentmods.dev/badge/commands/kimsanguine/hplan/harness-discover.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00051 | $0.04061 |
| Opus 5 | $0.00026 | $0.02031 |
| Sonnet 5 | $0.00010 | $0.00812 |
| Haiku 4.5 | $0.00005 | $0.00406 |
Grade A, and why
harness-discover 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 8d 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 — 356 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/harness-discover
기회 발굴 전체 워크플로우 — 아이디어 → 검증된 기회 + 비용 추정 + 빌드/바이 결정 + 가정 검증
Phase 0 — 가정 심문 (선택적 진입점)
아이디어가 있지만 "만들 가치가 있나"라는 가정이 불분명할 때 먼저 이 단계를 거치세요. Phase 1(Opportunity Mapping)은 가정이 어느 정도 정리된 상태에서 시작할 때 더 유효합니다.
진입 조건: 다음 중 하나라도 해당되면 Phase 0부터 시작합니다.
- 아이디어는 있지만 "왜 이게 필요한가"에 대한 가정이 불분명함
- 이전에 비슷한 아이디어를 시도했다가 실패한 경험이 있음
- "사람들이 쓸 것 같다"는 직관 외에 근거가 약함
Phase 0 실행: /socratic-question [아이디어]
socratic-question은 6가지 질문 유형(명료화·가정·증거·관점·함의·메타질문)으로
당신의 가정을 심문하고, "사고 검증 질문 세트" 1장을 만듭니다.
Phase 0 완료 기준: "누구의 / 무슨 문제를 / 무엇으로 / 무엇은 빼고 푼다" 가 한 문장으로 나오면
→ Phase 1으로 넘어가세요 (/harness-discover [아이디어])
Phase 0 생략: 가정이 이미 명확하거나 빠른 탐색이 목적이라면 바로 Phase 1부터 시작해도 됩니다.
Routing
Parse $ARGUMENTS for a --mode flag before proceeding:
| Flag | Execution |
|---|---|
--mode socratic |
Phase 0 — socratic-question 실행 후 종료 |
--mode opp |
Phase 1만 실행 후 종료 |
--mode assumptions |
Phase 2만 실행 후 종료 |
--mode cost |
Phase 3만 실행 후 종료 |
--mode build-or-buy |
Phase 4만 실행 후 종료 |
--mode validate |
Phase 5만 실행 후 종료 |
| 플래그 없음 | Phase 1 → 2 → Checkpoint → 3 → 4 전체 실행 (Phase 5 제외) |
The remaining text after the flag is the target idea. If no flag is present, the entire $ARGUMENTS string is the idea.
Instructions
You are running the Discovery Workflow for: $ARGUMENTS
Follow the Routing table above to determine which phases to execute. Each phase is self-contained and independently readable.
--mode socratic 선택 시: /socratic-question [아이디어]를 먼저 실행하세요. 사고 검증 질문 세트가 완성되면 /harness-discover [아이디어]로 돌아오세요. (socratic-question을 직접 실행하지 않고 안내만 합니다.)
Interview Discipline (Phase 전반에 적용)
왜: 인터뷰는 가설을 확인하는 자리가 아니라 모르는 것을 발견하는 자리입니다. 질문이 많을수록, AI가 생성한 페르소나를 쓸수록 발견의 질이 낮아집니다.
모든 Phase에서 고객·사용자와 대화할 때 아래 4가지 규칙을 따른다:
규칙 1 — 질문 1개씩 복합 질문 금지.
- ✅ "언제 마지막으로 이 문제를 겪었나요?"
- ❌ "이 문제가 언제, 어디서, 왜 발생하나요?"
규칙 2 — Multiple choice 우선 열린 질문보다 선택지를 먼저 제시한다.
- ✅ "A: 직접 해결 / B: 담당자에게 위임 / C: 그냥 넘어감 — 가장 가까운 것은?"
- ❌ "이 상황에서 어떻게 하시나요?"
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.
- 8d ago First seen · 356 lines · 51 tokens per session scan A c99ec47ed5dd
harness-discover is a command published in the GitHub repository kimsanguine/hplan (2 stars, last pushed 23d ago), licensed MIT. It adds 51 tokens to every session and 4,061 once invoked, about $0.0003 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).
start-design-plan
Start collaborative design process with brainstorming and planning.
start-implementation-plan
Create implementation plan from design document.
co-change
List files that historically co-change with a target (git log heuristic; COCHANGE-style).
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.