AgentSkillsCN

resume-handoff

通过分析交接文档中的上下文、验证当前状态并确保工作连续性，从交接文档中恢复工作。当您从前一阶段的交接文档中继续推进工作、接手另一位开发者留下的未竟之业、在中断后重新拾起工作——尤其是在交接内容已沉淀于思想、分享或交接文档之中时；当您在中断后重新找回工作脉络，或在已有交接文档的基础上快速上手现有工作时，可使用此技能。

SKILL.md
--- frontmatter
name: resume-handoff
description: Resume work from a handoff document by analyzing context, validating current state, and establishing continuity. Use when continuing work from a previous session with a handoff, picking up where another developer left off, resuming after a break when handoff exists in thoughts/shared/handoffs/, recovering context after interruption, or onboarding to existing work with handoff documentation.

Resume Handoff

Resume work by thoroughly analyzing a handoff document and validating the current state of the codebase to ensure continuity with previous work.

Initial Response

The skill adapts based on what's provided:

Path provided: Read handoff and linked plans/research FULLY (no limit/offset), then propose next action.

Ticket number (ENG-XXXX):

  1. •Run humanlayer thoughts sync to ensure local thoughts is up to date
  2. •Find most recent handoff in thoughts/shared/handoffs/ENG-XXXX/
  3. •Read handoff and all linked documents completely
  4. •Propose next action

No parameters: Ask for handoff path or ticket number.

Process

1. Read and Analyze Context

  • •Read the handoff completely (never use limit/offset)
  • •Extract critical information:
    • •Original tasks and stated objectives
    • •Recent changes made (files modified, features added)
    • •Key learnings and insights discovered
    • •Artifacts created or referenced (plans, research docs)
    • •Action items left for continuation
  • •Sub-agent recommendation: Spawn parallel research sub-agents to verify current state of critical areas
  • •Wait for ALL sub-agents to complete
  • •Read critical files from "Learnings" and "Recent changes" sections
  • •Cross-reference linked plans/research from thoughts/shared/plans/ or thoughts/shared/research/

2. Present Comprehensive Analysis

code
I've analyzed the handoff from [date] by [researcher]:

**Original Tasks:** [Status and verification of each]
**Key Learnings Validated:** [Which are still valid vs. changed]
**Recent Changes Status:** [Which are still present vs. missing/modified]
**Artifacts Reviewed:** [Key takeaways from each]
**Recommended Next Actions:** [Based on handoff + current state]
**Potential Issues:** [Conflicts, regressions, or divergences]

Proceed with [specific action], or adjust approach?

3. Create Action Plan

Use TodoWrite to create structured task list from:

  • •Action items explicitly listed in handoff
  • •Gaps or issues discovered during validation
  • •Logical next steps for completing the work

4. Begin Implementation

Once user approves:

  • •Reference handoff learnings throughout implementation
  • •Apply documented patterns and approaches
  • •Maintain awareness of previous context and decisions
  • •Create new handoff if significant new learnings emerge

Common Scenarios

  • •Clean Continuation: All documented changes present → proceed with action items
  • •Diverged Codebase: Changes occurred since handoff → reconcile, adapt plan
  • •Incomplete Work: Previous work only partially complete → finish in-progress first
  • •Stale Handoff: Significantly outdated → re-evaluate strategy, possibly create fresh research

Sub-agent Recommendations

  1. •Context Verification Phase: Spawn parallel research sub-agents to validate different aspects (file state, dependencies, test status)
  2. •Linked Document Analysis: If handoff references multiple plans/research, analyze each in parallel
  3. •Codebase State Validation: Use research sub-agents to verify assumptions about current code