Agent skill

Continue Task

by UniClipboard in UniClipboard/UniClipboard

Resume work from a $wrap state file. An agent skill from UniClipboard/UniClipboard.

AGPL-3.0Auto-check passedAgent Workflows

Install Continue Task

skills CLI
$ npx skills add UniClipboard/UniClipboard --skill continue-task -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install UniClipboard/UniClipboard continue-task --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/UniClipboard/UniClipboard.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/continue-task .claude/skills/continue-task && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
continue-task
GitHub stars
1.9k
Token cost
~1.7k tokens
SKILL.md length
505 words
Files
1
Skills in repo
24
Repo updated
First seen
Licence
AGPL-3.0

At a glance

Resume work from a $wrap state file. An agent skill from UniClipboard/UniClipboard.

  • Works in 6 steps: Find and read state file → Validate state freshness → Present context summary → …
  • Tasks that involve Session handoff
  • SKILL.md covers Purpose, When to trigger, Workflow and Handling edge cases, plus 2 more sections
  • Calls git, jq and gh; needs PROJECT_KEY

What it does

Continue Task is an agent skill from UniClipboard/UniClipboard. Resume work from a $wrap state file. Reads the structured active-task.json, presents a context summary, restores branch state, and lets you pick up exactly where the last session left off — no manual context rebuild.

Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Session handoff. The repository describes itself as: Real-time clipboard sync across all your devices — local-first, peer-to-peer, and end-to-end encrypted. No account. No cloud dependency. No central server. The licence is AGPL-3.0.

When your agent uses it

  • Tasks that involve Session handoff

Example prompts

  • “/continue-task”

Requirements

  • A credential in PROJECT_KEY

Workflow steps

6 steps, taken from the step headings in SKILL.md.

  1. Find and read state file
  2. Validate state freshness
  3. Present context summary
  4. Offer actions
  5. Load context references
  6. Clear state on completion

What it can do on your machine

Read from SKILL.md and the folder at commit 101ffb3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • git
    • jq
    • gh

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use git and gh, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • PROJECT_KEY

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Continue Task loads about 1.7k tokens when it runs. Until then it costs about 58 tokens; SKILL.md has 505 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~58
When it runs · the whole SKILL.md, loaded when a task matches
~1.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from UniClipboard/UniClipboard at commit 101ffb3, republished under its AGPL-3.0 licence (© UniClipboard). 505 words, ~1,741 tokens.

Download SKILL.mdSave it as .claude/skills/continue-task/SKILL.md (or your agent's skills folder).
name
continue-task
description
Resume work from a $wrap state file. Reads the structured active-task.json, presents a context summary, restores branch state, and lets you pick up exactly where the last session left off — no manual context rebuild.

continue-task

Purpose

The counterpart to $wrap. When a previous session ended with $wrap, this skill reads the structured state file and rebuilds context in seconds instead of minutes.

Before (observed pattern):

text
User: "继续上次的工作"
Agent: [reads git log, guesses what was happening, asks 3 questions, reads 5 files] — 3-8 minutes

After:

text
User: $continue-task
Agent: [reads active-task.json + context.md, presents summary] — 10 seconds
  "上次你在 active-clipboard-state 分支做 PR8 (pull store-only LWW),
   已完成 PR1-7, 下一步是加双节点 e2e 测试。继续?"

When to trigger

  • $continue-task — resume from state file
  • User says "继续上次的", "pick up where I left off", "resume the task"
  • At the start of a new session when the user references prior work

Workflow

Step 1 — Find and read state file
bash
PROJECT_KEY=$(pwd | sed 's|^/||; s|/|-|g')
STATE_DIR="${CODEX_HOME:-$HOME/.codex}/project-state/${PROJECT_KEY}"
STATE_FILE="${STATE_DIR}/active-task.json"
CONTEXT_FILE="${STATE_DIR}/active-task-context.md"

If the state file does not exist:

text
No active task found for this project.
Last $wrap state file not found at: <path>

Options:
  A) Check git log and branch state to reconstruct context
  B) Start fresh

If the state file exists, read both files.

Step 2 — Validate state freshness

Check if the state is stale:

bash
# Is the branch still current?
CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
STATE_BRANCH=$(jq -r '.branch' "$STATE_FILE")

# Has the branch moved since state was written?
STATE_TIME=$(jq -r '.created_at' "$STATE_FILE")

# Is there a PR and what's its status?
PR_NUMBER=$(jq -r '.pr.number // empty' "$STATE_FILE")
if [ -n "$PR_NUMBER" ]; then
  gh pr view "$PR_NUMBER" --json state --jq '.state' 2>/dev/null
fi

Staleness conditions:

  • State is more than 7 days old → warn, but still usable
  • Branch was deleted or merged → state is stale, report and offer to clean up
  • PR was closed/merged → task may be done, confirm with user
  • Current branch differs from state branch → ask which to use
Step 3 — Present context summary

Render a concise briefing:

text
📋 Active task: Implement ActiveClipboardState for cross-device LWW register
   Branch: active-clipboard-state
   PR: #1117 (open, 2 checks pending)
   Last wrapped: 2026-06-20 08:49

   ✅ Done:
     • PR1-PR6: core state machine, reconcile, sync
     • PR7: mobile cutover

   ➡️ Next:
     • PR8: pull store-only LWW starvation workaround
     • Add dual-node e2e test for PR8

   📁 Key context:
     • .planning/2026-06-19-issue-1017-active-clipboard.md
     • docs/architecture/active-clipboard-state.md

If there's an active debug state:

text
   🔍 Debug session in progress:
     Symptom: Windows restore → Mac sync delay ~3.4s
     
     Ruled out:
       ✗ #1017 logic bug (disproven: timing issue, not state bug)
       ✗ relay latency (disproven: direct conn established)
     
     Current hypothesis: iroh candidate address filtering not excluding Tailscale 100.x
     Next step: grep CIDR constants in node.rs, check filter function
Step 4 — Offer actions
text
What would you like to do?
  A) Continue from where we left off (start with next steps)
  B) Review the changes made so far (git diff, PR status)
  C) Start a different task (archive this state)
  D) Read the full context document first

On choice A:

  • If the branch differs from current, offer to switch: git switch <state_branch>
  • Read the context references listed in the state file
  • Begin working on the first item in task.next

On choice B:

bash
git log main..HEAD --oneline
git diff main...HEAD --stat
gh pr view <number> --json statusCheckRollup,reviews 2>/dev/null

On choice C:

  • Rename the state file to archived-task-<date>.json
  • Start fresh

On choice D:

  • Read and present the full active-task-context.md
Step 5 — Load context references

For each file in context_refs, check if it exists and is relevant:

bash
for ref in $(jq -r '.context_refs[]' "$STATE_FILE"); do
  if [ -f "$ref" ]; then
    echo "Reading: $ref"
    # Read the file to load context
  else
    echo "⚠️ Referenced file not found: $ref (may have been moved or deleted)"
  fi
done
Step 6 — Clear state on completion

When the resumed task is completed (user confirms or $wrap is called again with a new task), the old state is naturally overwritten.

If the user explicitly says "this task is done" or "任务完成了":

bash
# Archive the state
mv "$STATE_FILE" "${STATE_DIR}/archived-task-$(date +%Y%m%d-%H%M).json"
mv "$CONTEXT_FILE" "${STATE_DIR}/archived-task-$(date +%Y%m%d-%H%M)-context.md" 2>/dev/null

Handling edge cases

No state file, but user wants to continue

Fall back to git-based reconstruction:

bash
git log --oneline -10
git status
git branch -v
gh pr list --author @me --json number,title,headRefName --jq '.[]' 2>/dev/null

Present what can be inferred and ask the user to fill in gaps.

Show full SKILL.md (199 more words)Show less
State file from a different branch

If the user switched branches between sessions:

text
⚠️ State file references branch 'active-clipboard-state',
   but you're currently on 'main'.

   A) Switch to active-clipboard-state and continue
   B) Stay on main, ignore the state file
   C) Show me what's in the state file first
Multiple projects with state files

If the user works on multiple projects (uniclipboard, uc-website, uniclipboard-android), each has its own state file under its own project key. No conflict.

Parallel branches in the same project (worktrees)

Each worktree has a different pwd and therefore a different project key. State files are naturally isolated.

Interaction with other skills

SkillHow it interacts
$wrapCreates the state file that $continue-task reads
$handoffProduces prose-only document; $continue-task cannot consume it (but the context.md from $wrap is similar)
Codex resumeResumes a prior Codex conversation; $continue-task restores explicit task-level state across conversations
$pr-greenlightCan be the "next step" suggested by the state file
$error-diagnose-fixDebug state section helps this skill avoid retrying ruled-out hypotheses

Anti-patterns

  • Reading the state file but then ignoring it and asking the user "what are you working on?"
  • Loading all context references even when the user only wants to do a quick commit
  • Treating a stale state file (7+ days old) as current without warning
  • Not checking if the branch still exists before suggesting to switch
  • Spending 5 minutes "analyzing" the state when a 10-second summary suffices

© UniClipboard, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/continue-task of UniClipboard/UniClipboard.

Open the folder on GitHubat commit 101ffb3

Compare with similar skills

Continue Task next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Continue Task compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Continue Task this skillUniClipboard/UniClipboard1.9k—~1.7kAutomated safety check: PassAGPL-3.0
Orca CLIstablyai/orca87k2 repos~593Automated safety check: PassMIT
Coding Agent Session Findercode-yeongyu/oh-my-openagent70k1 repos~2.8kAutomated safety check: PassCustom licence
Beads Task Memorygastownhall/beads28k—~1.2kAutomated safety check: PassMIT
Session History Searchslopus/happy24k—~3.1kAutomated safety check: PassMIT
Paseo Agent Handoffgetpaseo/paseo20k1 repos~606Automated safety check: PassCustom licence

Similar skills

  • Orca CLI

    stablyai/orca

    Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser…

    87k GitHub starsUsed in 2 repos~593 tokens
    Agent WorkflowsAuto-check passed
  • Coding Agent Session Finder

    code-yeongyu/oh-my-openagent

    Finds, reads and reconstructs past coding-agent sessions across Codex, Claude, OpenCode, Senpi and many other local agent logs.

    70k GitHub starsUsed in 1 repo~2.8k tokens
    Agent WorkflowsAuto-check passed
  • Beads Task Memory

    gastownhall/beads

    Tracks multi-session work with dependencies in the bd issue tracker so the agent can find ready tasks and recover its context after conversation compaction.

    28k GitHub stars~1.2k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Searches past Claude Code, Codex and Cursor sessions and summarizes what was worked on, tried or decided, using extraction scripts instead of reading raw logs.

    24k GitHub stars~3.1k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed
  • Paseo Agent Handoff

    getpaseo/paseo

    Hands off the current task, including context, decisions and failed attempts, to a fresh agent through Paseo by writing a self-contained briefing prompt and launching that agent.

    20k GitHub starsUsed in 1 repo~606 tokens
    Agent WorkflowsAuto-check passed
  • Memori Long-Term Memory

    MemoriLabs/Memori

    Connects Claude Code to Memori Cloud for long-term memory, recalling stored context before substantive replies and saving new context afterward.

    17k GitHub stars~2k tokensUpdated 5 days ago
    Agent WorkflowsAuto-check: notes

More from UniClipboard/UniClipboard

All 24 skills in this repo
  • Beui

    UniClipboard/UniClipboard

    Pick and install beUI (@beui) animated React components from the shadcn registry.

    1.9k GitHub stars~1.8k tokensUpdated today
    Auto-check passed
  • Create PR

    UniClipboard/UniClipboard

    Push the current branch and open a GitHub pull request against main.

    1.9k GitHub stars~4.6k tokensUpdated today
    Auto-check passed
  • Design Audit

    UniClipboard/UniClipboard

    定期审计代码库的工程设计问题(高心智复杂度、单一真相源被破坏、catch-all 胖接口、死代码、散落魔法字面量、泄漏抽象、资源生命周期靠环形缓冲)与可优化点,范围限定为自上次审计以来的 git churn,每条发现都落到 file:line 并对照本项目自己的 VISION.md / 各级 AGENTS.md / memory…

    1.9k GitHub stars~554 tokensUpdated today
    Auto-check passed
  • Dual Side Debug

    UniClipboard/UniClipboard

    Inspect uniclipboard logs from BOTH the macOS host and the mounted Windows peer when debugging cross-platform sync, pairing, transfer, or daemon issues.

    1.9k GitHub stars~2.6k tokensUpdated today
    Auto-check passed
  • E2E Test Thinker

    UniClipboard/UniClipboard

    Analyze the current branch's diff against main and determine which changes are testable via CLI-based end-to-end tests.

    1.9k GitHub stars~1.7k tokensUpdated today
    Auto-check passed
  • iOS Log Diagnose

    UniClipboard/UniClipboard

    Drive the UniClipboard iOS app in a simulator and read its OSLog yourself to diagnose a mobile-sync bug, instead of asking the user to paste logs.

    1.9k GitHub stars~1.1k tokensUpdated today
    Auto-check passed

Categories

Questions about Continue Task

What does Continue Task do?

Resume work from a $wrap state file. An agent skill from UniClipboard/UniClipboard. Continue Task is an agent skill from UniClipboard/UniClipboard. Resume work from a $wrap state file.

When should I use Continue Task?

Continue Task fits situations like: tasks that involve Session handoff.

How do I install Continue Task in Claude Code?

Run `npx skills add UniClipboard/UniClipboard --skill continue-task -a claude-code`. Or copy the skill folder (.agents/skills/continue-task in UniClipboard/UniClipboard) into .claude/skills/continue-task in your project. Claude Code loads it when a task matches its description.

How do I install Continue Task in Codex?

Run `npx skills add UniClipboard/UniClipboard --skill continue-task -a codex`. Or copy the skill folder (.agents/skills/continue-task in UniClipboard/UniClipboard) into .agents/skills/continue-task in your project. Codex loads it when a task matches its description.

Can I use Continue Task in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add UniClipboard/UniClipboard --skill continue-task -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/continue-task, .gemini/skills/continue-task, .github/skills/continue-task and .opencode/skills/continue-task in your project.

What does Continue Task need to run?

Going by SKILL.md and its folder, Continue Task needs the command-line tools its instructions call (git, jq and gh) and credentials named PROJECT_KEY. Our summary lists: A credential in PROJECT_KEY.

Does Continue Task access the network?

SKILL.md contains no URLs. Its commands use git and gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Continue Task safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Continue Task use?

Continue Task is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Continue Task use?

About 1.7k tokens (SKILL.md is roughly 7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Continue Task?

Skills that share tags, products or a category with Continue Task: Orca CLI (stablyai/orca, 87k stars), Coding Agent Session Finder (code-yeongyu/oh-my-openagent, 70k stars), Beads Task Memory (gastownhall/beads, 28k stars) and Session History Search (slopus/happy, 24k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Continue Task?

UniClipboard (a GitHub organization) maintains it in UniClipboard/UniClipboard, which has 1,856 GitHub stars. The repository holds 24 skills in this directory. The repository was last updated on October 8, 2026.

Source: UniClipboard/UniClipboard on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.