Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add hxt9805/cadence/plugin install cadenceWrote 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/skills/hxt9805/cadence/cadence-handoff)<a href="https://agentmods.dev/skills/hxt9805/cadence/cadence-handoff"><img src="https://agentmods.dev/badge/skills/hxt9805/cadence/cadence-handoff/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/skills/hxt9805/cadence/cadence-handoff"><img src="https://agentmods.dev/badge/skills/hxt9805/cadence/cadence-handoff.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 | $0.00077 | $0.02078 |
| Opus 5 | $0.00039 | $0.01039 |
| Sonnet 5 | $0.00015 | $0.00416 |
| Haiku 4.5 | $0.00008 | $0.00208 |
Grade A, and why
cadence-handoff 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 11d 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 — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.
cadence-handoff(v0.4)
定位(v0.3 重构)
v0.3 前 handoff 承担"决策/待决/约束/TODO/incidents"五类提取 + 9 步写链;writer subagent context 随 session 大而膨胀。
v0.3/v0.4 后记/整/查三阶段覆盖这些职责:
- 记 阶段(α 流式)记录已承载"讨论内容持久化"(五类提取不再需要)
recall-retriever查 阶段(ρ)<500 tokens 契约负责跨 session 检索(注入历史不再需要)- 整 阶段(ε 整合)触发(含 handoff 兜底)自动收敛零散
- handoff 写书签前执行 fidelity sweep,补齐"对话已承接但尚未落档"的稳定语义
新本质:短小书签——提供"游标"(讨论到哪 / 卡在哪)、canonical 指针和 "soft context"(语气、未明说的注意事项)。详细内容不复制进 handoff,体量目标 15-30 行。
路径约定
所有 handoff 产物必须位于 <project-root>/cadence/.handoff/。不得在项目根创建或读取裸 .handoff/;发现时只提示迁移(见 cadence-resume/SKILL.md Step 1 前置检测)。
5 步流程(主 agent 执行,不派 writer subagent)
Step 1:fidelity sweep + 整 阶段(ε 整合)兜底
对账当前 session 已承接的持久语义与本 session 新增 streaming entry
→ 范围明确的漏记:按 project-discuss 保真规则自动 append,告知用户
→ 指向不明或有多个候选方案:询问用户,不得猜测
→ Glob cadence/streaming/*.md
→ 读每个文件 front-matter
→ 对每个 status: active 文件,检查 entry 保真等级、收尾语义、数量与时间新鲜度
→ High / 显式收尾 / 成熟主题 → 派 recall-consolidator(trigger_reason: handoff_sweep)
→ 接收 plan → 走两阶段写(复用 project-discuss 整 阶段流程)
→ 验证 source_entries 与 coverage 完全对应,否则保持 active
→ 把整合产出的 discussion doc 路径收集到 `produced[]`
只有不是当前 continuation、且未收尾的 Light/Standard 主题可因
entry <2 条 / 距今 <10 分钟 跳过。当前 continuation、High 或用户已明确收尾的
主题即使只有一条 entry 也必须尝试整合,确保新 handoff 至少有一个 canonical ref。
Partial success 语义:若至少一个主题整合成功 → consolidation.triggered: true,produced[] 仅含成功项;全部失败或跳过 → triggered: false,produced: []。
Step 2:扫对话产生 cursor + soft_context
主 agent 自己做(不派 subagent,因为新 schema 体量小,主 context 胜任):
cursor.last_discussed:session 最近一段讨论话题,一句话cursor.pending_questions:未决问题,0-3 个cursor.next_step:下次接着做什么,一句话(可空)soft_context.tone:对话语气(如"紧凑" / "探索中")(可空)soft_context.notes:特殊注意事项,0-N 条(可空)
Step 3:计算 content_hashes + continuation_refs
使用 git hash-object -w --stdin 跨平台计算(跨平台首选,cadence 项目已依赖 git):
INDEX_HASH=$(git hash-object cadence/_INDEX.md)
ACTIVE_HASH=$(git hash-object cadence/_ACTIVE.md)
备用路径(若 git 不可用):
- Windows(PowerShell):
(certutil -hashfile <path> SHA1 | Select-String -NotMatch "SHA1\|CertUtil").Line.Trim().Replace(" ","")— 取哈希行,剥离空格 - Windows(cmd):
for /f "tokens=1" %%i in ('certutil -hashfile ^<path^> SHA1 ^| findstr /v ":"') do @echo %%i— 取无冒号行第一列 - Linux:
sha1sum <path> | awk '{print $1}'— 取前 40 字符字段 - macOS:
shasum -a 1 <path> | awk '{print $1}'— Mac 默认无 sha1sum,用 shasum - Python fallback:
python -c "import hashlib,sys; print(hashlib.sha1(open(sys.argv[1],'rb').read()).hexdigest())" <path>(命令名:Windowspython/ macOS,Linuxpython3)
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 11d ago First seen · 144 lines · 77 tokens per session scan A 8e4333e409fe
cadence-handoff is a skill published in the GitHub repository hxt9805/cadence (5 stars, last pushed 1mo ago), licensed MIT. It adds 77 tokens to every session and 2,078 once invoked, about $0.0004 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 skills, from other repositories
flow-next-interview
In-depth Q&A to refine a spec, task, or spec file before building. Use when asked to flesh out, refine, or interrogate requirements.
flow-next-impl-review
Carmack-level implementation review of changes via the configured backend. Use when asked to review code or a diff in a flow-next repo.
flow-next-audit
Audit .flow/memory/ entries against current code and keep, update, consolidate, replace, delete, or harden each. Use when asked to audit memory or graduate a recurring lesson into a gate.
flow-next-capture
Synthesize the current conversation into a flow-next spec with read-back gating. Use when asked to capture this as a spec.
flow-next-drive
Drive any UI surface like a real user - a web app, a Chromium-backed desktop app (Electron / WebView2, reached over CDP), or a genuinely native app (macOS AppKit/SwiftUI, or a non-CDP webview) reached via the Cua Driver / Computer Use. Detects the surface, picks the best available driver, degrades gracefully. Use to…
flow-next-make-pr
Open a PR with a cognitive-aid body rendered from flow-next spec state via gh. Use whenever asked to make or open a PR in a flow-next repo.