Install
$ agentstack add skill-christophezhao-vibe-coding-skills-session-handoff ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Session Handoff
Purpose
Use this skill to offload the current session into a continuation pack before opening a new window for the same unfinished task. The primary output is one Markdown handoff file under docs/session-handoff/ that helps the next window quickly recover task state, blockers, next action, and relevant lessons without manually rebuilding context.
Activation Cues
Activate when any condition is true:
- User explicitly asks to整理当前 session, 做 handoff, 做 context offload, or prepare a new window to continue the same task.
- The current task is unfinished and the user wants a session-level summary for next-window continuation.
- User asks to summarize what this window has done, what remains, and what the next window should read first.
Do not activate when any condition is true:
- The request is only to create reusable lessons or an experience card.
- The request is only to update plan status or lifecycle governance.
- The request is only to preserve long-term project memory without a window handoff need.
- The request is full transcript archival or raw chat export.
Scope Limit
This skill manages session-level continuation packs. This skill does:
- summarize current-session progress for next-window continuation
- aggregate high-signal refs from plans, memory, experience, and process docs
- preserve blockers, next actions, and avoid-repeat guidance
- point the next window to the right execution entrypoint before coding when direct coding would be unsafe
This skill does not:
- own plan lifecycle status
- own reusable experience-card persistence
- own project memory truth
- archive full chat transcripts
- auto-write back to plan, memory, or experience docs in MVP
- author or refresh implementation plans itself
Ownership Boundary
sdd-plan-maintainerowns plan lifecycle and plan status truth underdocs/plans/.layered-project-memoryowns continuity facts underdocs/memory/.experience-captureowns reusable cross-task experience underdocs/experience/.session-handoffowns session-level aggregation for next-window continuation.
Boundary rule:
- Read canonical artifact surfaces under
docs/rather than depending on a specific skill name or output. - Treat plan, memory, and experience docs as source of truth; the handoff pack is only a pointer-first aggregation layer.
- If a next-window plan review or plan creation should happen before coding, point to the right owner and current entrypoint, but do not write the plan content here.
Script Decision
Use the skill-bundled scripts/smoke.sh for structure and contract checks only. No business runtime script is required in MVP.
Workflow Contract
- Confirm the request is a current-window offload for same-task continuation.
- Collect related plan, memory, experience, and process refs from
docs/and other referenced evidence files. - Write one handoff pack to
docs/session-handoff/SHO-YYYYMMDD-NNNN.md. - Keep the pack pointer-first and scoped to the current session, even when the session spans multiple subtasks.
- Ensure the pack includes fixed sections:
Session IDGenerated AtOffload ReasonSession GoalOverviewRelated RefsSubtask SnapshotsAvoid RepeatNext Window BootSource of Truth
- Limit
Avoid Repeatto 1-3 high-signal items. - New-window consumption order must be stated clearly:
- read
Offload ReasonandOverview - read
Subtask Snapshots - read related plan refs
- read related memory refs
- read related experience refs and
Avoid Repeat - then start new planning
- When direct coding would be unsafe, add one concise
Before codinghint insideNext Window Bootthat points to the current execution entrypoint or the next planning owner.
Keep it pointer-first; do not turn the handoff into a plan or execution contract.
Hard Rules
- MVP is explicit-trigger only; no
suggest-once. - Output exactly one Markdown continuation pack per trigger.
- Do not create
.jsonsidecars orSESSION_INDEX.jsonin MVP. - Do not auto-refresh or mutate plan, memory, experience, or process docs in MVP.
- Keep references repo-relative whenever possible.
- Do not copy large logs or long document excerpts; summarize and point.
- If the current session invalidates an older plan, mark that plan as historical or reference-only in the pack and point to the new diagnostic entry, but do not mutate plan lifecycle state in this skill.
- Do not author plan steps, acceptance criteria, or execution contracts inside the handoff pack; only point to the next planning owner when needed.
Resource Map
- Read
references/positioning-boundary.mdfor scope and layer split. - Read
references/trigger-rules.mdfor trigger and non-trigger rules. - Read
references/canonical-surfaces.mdfor artifact-surface reading rules. - Read
references/process-protocol.mdfor execution flow and handoff order. - Read
references/output-contract.mdfor Markdown section contract. - Read
references/resume-entrypoint-guidance.mdwhen the next window may need a plan review or a plan refresh before coding. - Read
references/regression-cases.mdfor trigger-boundary regression checks. - Read
references/examples/README.mdfor examples and non-examples. - Use the skill-bundled
scripts/smoke.shfor deterministic smoke checks.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: ChristopheZhao
- Source: ChristopheZhao/vibe-coding-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.