Skip to content
Agentmart
Submit a tool

binder-replay

当需要回放已记录的 binder 设计 campaign、重现每一步"当时知道什么、决定了什么"时使用;对应 openbinder replay(--fixture/--events/--source)。典型触发语:"帮我回放一下 IL-7Ra 的 binder campaign""回放那个失败的…

Skill by Hylouis233

IndexedMIT
Write a review

Chalk Labs

Need help deploying binder-replay?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/hylouis233-binder-replay

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

当需要回放已记录的 binder 设计 campaign、重现每一步"当时知道什么、决定了什么"时使用;对应 openbinder replay(--fixture/--events/--source)。典型触发语:"帮我回放一下 IL-7Ra 的 binder campaign""回放那个失败的 campaign""看看当时每一步都知道了什么"。同义场景:campaign replay、事件流回放、replay report、gap 分析、时间线重建。

Use cases: documents, marketing

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/binder-replay

What's inside

Skill name
binder-replay
Description
当需要回放已记录的 binder 设计 campaign、重现每一步"当时知道什么、决定了什么"时使用;对应 openbinder replay(--fixture/--events/--source)。典型触发语:"帮我回放一下 IL-7Ra 的 binder campaign""回放那个失败的 campaign""看看当时每一步都知道了什么"。同义场景:campaign replay、事件流回放、replay report、gap 分析、时间线重建。
License
MIT
Metadata
last_reviewed: 2026-08-20
Instructions
76 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# binder-replay:campaign 事件流回放

## 目的

回放把公开的 campaign 过程记录(per-design provenance)归一化成统一事件流,重建 campaign 状态与逐步时间线:每一步当时掌握了什么证据、做了什么决定、哪里是缺口。成功**和**失败的 campaign 都回放——失败 campaign 的 gap 往往比成功的时间线更有信息量。全程不执行任何蛋白模型、不调任何 LLM。

## 前置检查

1. 确认输入三选一:
   - `--fixture {synthetic-success, synthetic-failure}`:内置合成 fixture,完全离线;
   - `--events <campaign-events.jsonl> --campaign <campaign_id>`:自带事件流,离线;
   - `--source anthropic-claude-binder-v1 --target <target> [--campaign <id>]`:从已登记源按下载策略逐条拉取 per-design `provenance.json`,**需要网络**。
2. `--source` 模式额外检查:源已 lock;未知 target/campaign 会 exit 2 并列出 API 实际提供的可选项——把列表原样给用户,不猜。
3. 产物落盘:`output/binder-replay/<slug>/<timestamp>/`;同一源多 campaign 时每个 campaign 一个子目录。

## 1 · 执行回放

```bash
# 离线 fixture
openbinder replay --fixture synthetic-failure --out output/binder-replay/<slug>/<timestamp>/
# 自带事件流
openbinder replay --events events.jsonl --campaign <campaign_id> --out <dir>
# 真实源(网络,策略门控)
openbinder replay --source anthropic-claude-binder-v1 --target <target> --out <dir>
```

产物(`replay-report.md` + `replay-state.json` + provenance 记录):

- 报告四节:Summary / Timeline / Decisions / **Gaps**;
- `--source` 模式的报告追加 `## Source availability` 一节,如实列出 design_summary / wetlab_labels / docs / manifests 四类表的可用状态;
- 每次运行登记 provenance(campaign、status、gaps 数)。

## 2 · gap 语义(报告的核心纪律)

- 以下情况一律记为 gap:**跳过的规范阶段、不归属于任何阶段的事件、未 finalize 的事件流**;
- gap 是"记录里没有什么",**永不插值、永不脑补**——不把相邻步骤的内容缝进缺口;
- 声明了但缺失的字段以 `None` + `missing_fields` 呈现;CLI 会打印缺失计数,转述时不得抹掉。

## 3 · 确定性

- 报告与状态文件**逐字节确定**:相同输入重跑产出 byte-identical  artifact;报告不含任何墙钟时间(`ts` 只存在于 `.openscience/provenance.jsonl`);
- 因此 replay 产物可以直接进证据胶囊与对比报告,diff 即审计。

## 4 · 失败语义

- 未知 campaign / 缺文件 / 未知 target:exit 2,错误信息列出实际可选项;
- 下载被策略拦截(`blocked`):exit 2 + 结构化原因,是纪律生效;
- 源不可达(`unavailable`):exit 2,是 source gap,**不等于 campaign 不存在**;
- 回放本身不需要审批:五个人工 stage gate 不含 replay,事件回放永不阻塞等待批准。

## 输出模板

```markdown
## 回放结果:<campaign_id>

- 状态:<status>;事件数 / 设计数
- 产物:replay-report.md / replay-state.json 路径
- 时间线要点:<3-6 条关键步骤与决定>
- Gaps:<逐条列出;无则写"无已知缺口">
- Source availability(--source 时):<四类表状态>
- provenance:已登记(tool=openbinder:replay)
```

## 本技能不做什么

- 不执行蛋白模型:回放的是**记录**,candidate 生成与打分永远不在本仓库重跑。
- 不调 LLM:replay 引擎是纯规则重建;要模型决策请用 binder-agent-eval。
- 不填补 gap:缺口如实报告,"没记录"永远不写成"没发生"或"失败了"。
- 不承诺复现:没有重跑蛋白模型,generation campaign 永远不能叫 reproduced(docs/scope.md §8)。

## 收尾与下一步

- 向用户报告 campaign 状态、时间线要点与全部 gap;有 missing_fields 时明确计数。
- 下一步通常是 binder-ranking-audit(审计该 campaign 的排序)或 binder-agent-eval(在相同阶段输入上评测模型决策)。

From Hylouis233/workbuddy-skills, MIT.

More from Hylouis233/workbuddy-skills

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 23, 2026 (7 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from Hylouis233

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get binder-replay running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying binder-replay?

CallGet help
Cite this page

binder-replay skill. Agent Mart. https://agentmart.dev/skills/hylouis233-binder-replay (last updated 2026-10-24)