- Home
- Skills
- Agents & Orchestration
- Task Handoff
Task Handoff
Move unfinished coding work into the next session with completed work, evidence, constraints and the first unfinished step intact.
Free
Works with the AI Agents you already use
Task Handoff
Example session with this skill installed
Prepare a handoff of this skill-product task. The parser demo and helper tests are complete; marketplace publication is still pending. Preserve the results, evidence and current limits. Do not publish or rerun the checks just to write the handoff.
- Read your context and instructions
- Compiled the task handoff
Authoring-session walkthrough recorded on 2026-11-11, not an independent-agent benchmark. The following is the handoff from that session.
Goal and scope
Prepare two G-Code-derived skill products: paid Proof of Fix and free Task Handoff. Current work is local packaging and review. No account setup, external submission or email sending has occurred.
Where work stands
Both skill folders are under skills/. Seller drafts are under seller/. The executable example is under demo/proof-of-fix/. Original G-Code source was inspected as reference and was not edited for this task.
Completed
The helper’s 28 unit tests passed on Python 3.9.6 and 3.13.14 on macOS; these are the same 28 cases on two runtimes, not 56 different cases. Both SKILL.md files passed the skill-creator structural validator. In the parser fixture, six unchanged tests produced four failures before the fix and six passes after it. The initial failure log remains at demo/proof-of-fix/evidence/before-fix.log. The final log and manifest are in that same evidence directory. A deliberate source edit made the manifest stale; the exact tested bytes were then restored. No additional test run was needed merely to summarize those known results.
Facts needed next
Agensi’s current documentation says buyers receive skills through its connector and scripts run temporarily. Keep the two standalone upload ZIPs separate from the seller kit. The proposed paid price is USD 9 once; Task Handoff is proposed free. These are proposals, not live prices.
Unfinished and uncertain
No Agensi upload, platform security scan, approval, buyer delivery or sale has occurred. Name availability and the seller’s payout setup are unverified. No separate cold-agent behavior evaluation or Windows/Linux trial has run. The source hash helper checks consistency, not test authenticity.
Resume here
Review the finished seller kit and proposed price with the owner. When publication is authorized, use the owner’s connected account, inspect its payout readiness, upload only the intended skill ZIP, and verify the listing’s actual status after submission. Pending review is not live. Do not repeat completed tests unless the packaged source or a relevant environment changes.
Connects securely to your tools. The creator never sees your data.
About this skill
Long coding tasks often lose their place when a conversation ends. Task Handoff turns the available task record into a concise continuation note: the goal, current workspace, completed changes and checks, facts needed next, unresolved work and the next action.
It distinguishes a started command from a completed one, retains known results when a transcript is shortened, and keeps earlier completed work unless later evidence invalidates it. On resume, the agent checks the specific facts that may have drifted rather than redoing the entire investigation.
The output follows the user’s language and requested destination. This is a handoff-writing and resume workflow; it does not monitor context automatically, create chats, store memory, control processes or guarantee token savings.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Cannot recover missing conversation history or authenticate inherited test claims. Does not automatically create chats, monitor token limits, store durable memory or control processes. No guaranteed token savings. Only an authoring-session walkthrough has been performed; independent agent and Agensi buyer-delivery testing remain unverified.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
This skill has been used 1 time.
Trust & safety
Security scan
Security scan passed 4 days ago
- Free with an Agensi account
Nothing to install. It runs in your AI agent.
Use the relevant conversation or task record. File/status access is optional for freshness checks. No software installation is required.
Permissions
Read the supplied task context and specifically relevant files or status. Write a handoff only when a saved file is requested or expected by the workflow. No network service, credential access, remote messaging or automatic task execution is required.