searchflow

A research workflow command that investigates a question, gathers sources, checks their quality, and produces a structured report. Its description is partly in Korean, so the exact user-facing process cannot be fully confirmed here.

In plain words
What is it for?
Use it for structured web, news, market, academic, and technical research with source records and a final report.
Why use it?
It is intended to make research depth and source checking follow a defined process instead of relying on an informal search.

Command

Part of the searchflow plugin — 1 skill, 2 commands, 1 agent, 1 MCP server shipped together

Install

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.

agentmods
npx agentmods add commands/treylom/deep-research/searchflow
Clone the repo
git clone --depth 1 https://github.com/treylom/deep-research

Or install searchflow, the plugin that ships this one along with the rest of its 1 skill, 2 commands, 1 agent, 1 MCP server.

Per session 38 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 744 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00038 $0.00744
Opus 5 $0.00019 $0.00372
Sonnet 5 $0.00008 $0.00149
Haiku 4.5 $0.00004 $0.00074

Measured 2d ago against content hash 388290d9e6e2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

searchflow 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 2d 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.

commands/searchflow.md · 51 lines

What it actually says

/searchflow

$ARGUMENTS 를 조사한다.

실행

skills/searchflow/SKILL.md 를 정본으로 그 공정을 따른다. 요지:

  1. P0 스코핑 — 유형 판정(8종 + 복합, references/frames.md) → 프레임 축 확정 → 깊이 자동 라우팅 → 묶음 질문 1회(마감 시한 포함) → 권한 preflight → run_id 발급
  2. P1 스폰 — 프레임 축마다 조사 단위 1개 (하네스별 분기 = SKILL.md §3)
  3. P2 수집 — 워커는 근거 + SELF-REPORT(Y/N + 근거 1줄)만 반환. 등급 부여 ❌
  4. P3 채점 — 리드가 등급 부여 → out/sources.jsonl 기록(single-writer) → 축별 채점 → 문턱 대조
  5. P4 게이트 — 통과 시 종료 / 미달 시 최저 축만 재조사, 최대 2라운드

시작 시 1회

node skills/searchflow/scripts/env-detect.mjs        # 인자 없이 호출 = exit 0 고정, stdout 1줄 JSON

결과는 리드만 소비한다. enhanced 부재 = 에러가 아니라 core 강등 + 보고서 격하 라벨. multi_agent_api 는 참고값이며 multi-agent 경로의 최종 판정은 리드가 자기 tool 목록으로 한다.

산출

  • out/report.mdreferences/report-contract.md 칸 목록 전건
  • out/sources.jsonl — schema v1

기록 후 검증:

node skills/searchflow/scripts/grade-ledger.mjs out/sources.jsonl
# exit 0=PASS · 1=위반(수리) · 2=스크립트 오류(통과 취급 ❌ — 사유 기록 후 core 진행)

하지 말 것

  • 사용자에게 깊이를 고르게 하지 말 것 — 시스템이 라우팅한다(SKILL.md §2)
  • 워커 프롬프트에 점수·가중치·문턱을 넣지 말 것 — 재조사 지시는 "이 축이 약하다"까지
  • robots.txt/ToS/페이월/CAPTCHA 우회 금지 — 미도달로 기록(references/acquisition.md §2)
  • 원장을 워커가 append 하지 말 것 — 리드 single-writer

/deep-research 사용자

구 명령은 이 명령으로 위임되는 얇은 리다이렉트로 존치한다. 기존 run 원장은 read-only 로 보존되며 신규 run 만 schema v1 로 기록된다 — 구 데이터를 변환하지 않는다(무손실 보존 + 이중 계약 회피). 기존 run 의 재개(resume)는 지원하지 않으니 신규 run 으로 재실행한다.

Changes

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.

  1. 2d ago First seen · 51 lines · 38 tokens per session scan A 388290d9e6e2

Subscribe to this mod's changes

searchflow is a command published in the GitHub repository treylom/deep-research (2 stars, last pushed 1mo ago), licensed MIT. It adds 38 tokens to every session and 744 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.