返回技能列表

read-continue-here

pjt222
更新于 Yesterday
3 次查看
17
2
17
在 GitHub 上查看
设计ai

关于

This Claude Skill automatically resumes work from interrupted sessions by reading a CONTINUE_HERE.md handoff file at session start. It handles file detection, freshness checking, structured parsing, and user confirmation before cleaning up. Use it when bootstrapping after interruptions or setting up auto-continuation workflows.

快速安装

Claude Code

推荐
主要方式
npx skills add pjt222/agent-almanac -a claude-code
插件命令备选方式
/plugin add https://github.com/pjt222/agent-almanac
Git 克隆备选方式
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/read-continue-here

在 Claude Code 中复制并粘贴此命令以安装该技能

技能文档

Read Continue Here

Read structured handoff file. Resume work from where prior session stopped.

When Use

  • Start new session. CONTINUE_HERE.md sits in project root
  • After SessionStart hook injects handoff context
  • Bootstrap identity. Detect prior session artifacts
  • Set up auto continuation detection for project (one-time infra)

Inputs

  • Required: Project directory (default: cwd)
  • Optional: Configure infra (SessionStart hook + CLAUDE.md instruction)
  • Optional: Delete file after consume (default: yes)

Steps

Step 1: Detect and Read Handoff File

Check for CONTINUE_HERE.md in project root.

ls -la CONTINUE_HERE.md 2>/dev/null

Absent? Exit gracefully — nothing to continue from.

Present? Read contents. Parse 5 sections: Objective, Completed, In Progress, Next Steps, Context. Pull timestamp and branch from header line.

Got: File read. Sections parsed into clear mental model of prior session state.

If fail: File exists but malformed (missing sections, empty)? Treat as partial signal — pull what is there. Note gaps to user.

Step 2: Check Freshness

Compare file timestamp against current time.

# File modification time
stat -c '%Y' CONTINUE_HERE.md 2>/dev/null || stat -f '%m' CONTINUE_HERE.md
# Current time
date +%s

Classify freshness:

  • Fresh (< 24 hours, same branch): safe to act on direct
  • Stale (> 24 hours or different branch): flag to user before proceed
  • Superseded (new commits exist after handoff timestamp): someone worked on project since handoff

Check branch alignment.

git branch --show-current
git log --oneline --since="$(stat -c '%Y' CONTINUE_HERE.md | xargs -I{} date -d @{} --iso-8601=seconds)" 2>/dev/null

Got: Freshness verdict (fresh, stale, superseded) with evidence.

If fail: Not in git repo? Skip branch and commit checks. Use timestamp in file header alone.

Step 3: Summarize and Confirm Resume

Show handoff state to user concise:

  • "Prior session objective: [Objective]"
  • "Completed: [summary]"
  • "In progress: [summary]"
  • "Proposed next action: [Next Steps item 1]"

Stale or superseded? Show evidence. Ask whether proceed with handoff or start fresh.

Next Steps items tagged **[USER]**? Surface them explicit — they need user decision before work proceed.

Got: User confirms resume plan, maybe with tweaks. Agent has clear mandate for next action.

If fail: User says "start fresh" or "ignore that file"? Acknowledge. Proceed without handoff context. Offer delete file to prevent future confusion.

Step 4: Act on Handoff

Start work from Next Steps item 1 (or where user pointed):

  • Reference In Progress items to grasp partial state
  • Use Context section to avoid retry of failed approaches
  • Treat Completed items as done — no re-verify unless user asks

Got: Agent productive on right task, informed by handoff file.

If fail: Next Steps ambiguous or In Progress unclear? Ask user for clarity. No guess.

Step 5: Clean Up

Handoff consumed and work underway? Delete CONTINUE_HERE.md.

rm CONTINUE_HERE.md

Stale handoff files cause confusion in future sessions.

Got: File removed. Project root clean.

If fail: User wants keep file (e.g., as reference during session)? Leave it but note: must delete before session end. Else next session re-consumes.

Step 6: Configure SessionStart Hook (Optional)

Not yet configured? Set up auto reading of CONTINUE_HERE.md on session start.

Create hook script:

mkdir -p ~/.claude/hooks/continue-here

cat > ~/.claude/hooks/continue-here/read-continuation.sh << 'SCRIPT'
#!/bin/bash
# SessionStart hook: inject CONTINUE_HERE.md into session context
# OS-aware: works on native Linux, WSL, macOS, and Windows (Git Bash/MSYS)
set -uo pipefail

# --- Platform detection ---
detect_platform() {
  case "$(uname -s)" in
    Darwin) echo "mac" ;;
    Linux)
      if grep -qi microsoft /proc/version 2>/dev/null; then
        echo "wsl"
      else
        echo "linux"
      fi ;;
    MINGW*|MSYS*|CYGWIN*) echo "windows" ;;
    *) echo "unknown" ;;
  esac
}
PLATFORM=${PLATFORM:-$(detect_platform)}

CONTINUE_FILE="$PWD/CONTINUE_HERE.md"

if [ ! -f "$CONTINUE_FILE" ]; then
  exit 0
fi

# Strip CRLF (files on NTFS often have Windows line endings)
CONTENT=$(sed 's/\r$//' "$CONTINUE_FILE")

# JSON-escape: prefer jq, fall back to portable awk
if command -v jq >/dev/null 2>&1; then
  ESCAPED=$(printf '%s' "$CONTENT" | jq -Rsa .)
else
  ESCAPED=$(printf '%s' "$CONTENT" | awk '
    BEGIN { ORS=""; print "\"" }
    {
      gsub(/\\/, "\\\\")
      gsub(/"/, "\\\"")
      gsub(/\t/, "\\t")
      if (NR > 1) print "\\n"
      print
    }
    END { print "\"" }
  ')
fi

cat << EOF
{"hookSpecificOutput":{"sessionStartContext":{"additionalContext":$ESCAPED}}}
EOF
SCRIPT

chmod +x ~/.claude/hooks/continue-here/read-continuation.sh

Add to ~/.claude/settings.json in SessionStart hooks array.

{
  "type": "command",
  "command": "~/.claude/hooks/continue-here/read-continuation.sh",
  "timeout": 5
}

Got: Hook script exists, executable, registered in settings.json. Next session start: if CONTINUE_HERE.md exists, content injects into session context.

If fail: Verify settings.json valid JSON after edit. Test hook manual: cd /your/project && ~/.claude/hooks/continue-here/read-continuation.sh. Script falls back to awk if jq not installed — jq recommended, not required.

Step 7: Add CLAUDE.md Instruction (Optional)

Add brief instruction to project CLAUDE.md so Claude knows file purpose.

## Session Continuity

If `CONTINUE_HERE.md` exists in the project root, read it at session start. It contains a structured handoff from a prior session: objective, completed work, in-progress state, next steps, and context. Act on it — acknowledge the continuation, summarize prior state, and propose resuming from the Next Steps section. If the file is older than 24 hours, flag this to the user before proceeding. After the handoff is consumed, the file can be deleted.

Got: CLAUDE.md has instruction. Future sessions read and act on CONTINUE_HERE.md even when SessionStart hook not configured.

If fail: CLAUDE.md absent? Create with this section only. File too long? Add instruction near top where it not get truncated.

Checks

  • CONTINUE_HERE.md detected (or absence handled gracefully)
  • Freshness checked (timestamp, branch, post-handoff commits)
  • Resume plan shown to user, confirmed
  • Work began from correct Next Steps item
  • File cleaned up after consume
  • (Optional) SessionStart hook script exists, executable
  • (Optional) CLAUDE.md has session continuity instruction

Pitfalls

  • Act without confirm: Always show resume plan to user. They may have changed mind on what to work on, even when file fresh.
  • Trust stale files blind: Handoff file older than 24 hours or from different branch = suggestion, not mandate. Always check freshness.
  • Ignore Context section: Most valuable part = failed approaches. Skip it = retry dead ends.
  • Forget cleanup: Leave CONTINUE_HERE.md after consume = next session re-acts on it = confusion.
  • Treat Completed as unverified: Unless user asks specific, no re-do completed work. Trust prior session.

See Also

  • write-continue-here — complement: write handoff file at session end
  • bootstrap-agent-identity — full identity reconstruction; uses handoff detect as one heuristic
  • manage-memory — durable cross-session knowledge (complement to this ephemeral handoff)
  • write-claude-md — project instructions where optional continuity guidance lives

GitHub 仓库

pjt222/agent-almanac
路径: i18n/caveman/skills/read-continue-here
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

相关推荐技能

executing-plans

设计

该Skill用于当开发者提供完整实施计划时,以受控批次方式执行代码实现。它会先审阅计划并提出疑问,然后分批次执行任务(默认每批3个任务),并在批次间暂停等待审查。关键特性包括分批次执行、内置检查点和架构师审查机制,确保复杂系统实现的可控性。

查看技能

requesting-code-review

设计

该Skill可在完成任务、实现主要功能或合并代码前自动调度代码审查子代理,确保实现符合需求和计划。它支持通过指定git SHA范围进行精准的代码变更审查,帮助开发者在关键节点及时发现潜在问题。核心原则是"早审查、勤审查",适用于开发流程的各个关键阶段。

查看技能

connect-mcp-server

设计

这个Skill指导开发者如何将MCP服务器连接到Claude Code,支持HTTP、stdio和SSE三种传输协议。它涵盖了从安装配置到认证安全的完整流程,适用于集成GitHub、Notion、数据库等外部服务。当开发者需要添加集成、配置外部工具或提及MCP相关功能时,这个Skill能提供实用的操作指南。

查看技能

web-cli-teleport

设计

该Skill帮助开发者根据任务特性选择Claude Code的Web或CLI界面,并指导如何在两种环境间无缝迁移会话。它能分析任务复杂度、迭代需求等要素,推荐最优工作界面和工作流。关键特性包括会话状态管理、环境切换指导和上下文优化建议。

查看技能