Handover Skills 1.4.0

Adds the local-first handoff Agent Skill for switching sessions, models, agents, or owners without losing working state. The skill creates or resumes a concise HANDOFF.md without requiring an account, then optionally publishes it privately through Handover MCP or the CLI with identity and read-back verification. Existing skill slugs remain unchanged.

Handover Research and Evidence Index v1.0

A versioned, first-party evidence package for durable AI agent handoffs and human-agent continuation.

Included:

• AI Handoff Continuity Benchmark finding and source artifacts
• Reporter to Handover production migration field report
• Handoff Continuity Record v1.0
• Runnable human-agent continuation demo
• JSON Schema, CFF, and BibTeX citation metadata

Every item declares provenance, source links, and material limitations. This release is an inspectable first-party record, not independent validation.

Canonical human index: https://handover.sh/research
Agent-readable index: https://handover.sh/research.json

Handover CLI v0.1.4

Handover CLI v0.1.4 connects GitHub and npm discovery to the private AI handoff prompt generator before users install or configure an agent.

• Adds tracked GitHub and npm paths to the browser-only handoff prompt generator.
• Keeps generated draft content out of analytics and server-side draft storage.
• Adds executable-to-manifest version parity testing.
• Publishes through npm trusted publishing with SLSA provenance.

Install:

npm install --global handover-sh@0.1.4
handover --version

Try the no-account generator: https://handover.sh/tools/ai-handoff-prompt-generator?utm_source=github&utm_medium=release&utm_campaign=ai_handoff_prompt

Handover Skills 1.3.0

Adds the open Handoff Continuity Record v1 assets and the installable handover-record Agent Skill for creating, validating, publishing, and verifying portable handoff state.

Handover Agent Skills v1.2.0

Adds handover-test-continuity, a release-grade workflow for qualifying a complete Handover MCP handoff across separate authenticated identities.

The skill verifies multi-file read-back, revision-anchored review, optimistic-concurrency correction, resolution against the correcting revision, fresh-successor recovery, and denied, read-only, stale-revision, and revoked-credential paths. It reports connectivity, workflow completion, and continuity separately so a successful connection is never mistaken for a complete handoff.

Install the collection:

npx skills add 44-pixels/handover-mcp

Install only the continuity test:

npx skills add 44-pixels/handover-mcp --skill handover-test-continuity

Catalog: https://skills.handover.sh/
Guide: https://handover.sh/guides/end-to-end-mcp-agent-handoff?utm_source=github&utm_medium=release&utm_campaign=handover_test_continuity

Handover Public Continuation Demo

A complete, inspectable handoff that moves one reporting task across two agents and a human reviewer.

The demo includes Markdown, SQL, JSON, an image, three attributable revisions, exact human review evidence, a resolved note, and one bounded next action. No account is required to inspect it.

• Inspect the demo: https://handover.sh/demo?utm_source=github&utm_medium=release&utm_campaign=public_continuation_demo
• Continue it in a private workspace: https://handover.sh/app?start=demo&utm_source=github&utm_medium=release&utm_campaign=public_continuation_demo
• Read the agent record: https://handover.sh/demo.json?utm_source=github&utm_medium=release&utm_campaign=public_continuation_demo
• Review the workflow and limits: https://github.com/44-pixels/handover-mcp/blob/main/DEMO.md
• Connect MCP or the CLI: https://handover.sh/install?utm_source=github&utm_medium=release&utm_campaign=public_continuation_demo

Following the continuation link opens a prefilled draft after sign-in. It never creates private context until the user explicitly selects Create.

Built to be continued.

AI Handoff Continuity Benchmark v1.0 pilot

An open, reproducible pilot for testing whether an AI model recovers continuation-critical state from a conversation transcript, compressed memory, or structured handoff.

Pilot result across OpenAI gpt-5.6-terra and Anthropic claude-sonnet-5:

• Structured handoff: 79.45
• Conversation transcript: 76.67
• Compressed memory: 45.00

This is a small authored pilot, not a model leaderboard. One run per system does not estimate variance, the conditions retain different amounts of detail, and schema adherence affected scores.

Inspect and reproduce:

• Method and limitations: https://handover.sh/benchmark
• Dataset and runner: https://github.com/44-pixels/handover-mcp/tree/dd62884/benchmark/v1
• Raw responses: https://github.com/44-pixels/handover-mcp/tree/dd62884/benchmark/v1/results/2026-08-04/raw
• CFF: https://handover.sh/benchmark/v1/CITATION.cff
• BibTeX: https://handover.sh/benchmark/v1/citation.bib
• CSV: https://handover.sh/benchmark/v1/results/2026-08-04/summary.csv

Dataset license: CC BY 4.0. Runner and repository code: MIT.

Handover CLI v0.1.0

The first public Handover CLI release.

Install:

npm install --global handover-sh
handover login
handover whoami

Use it to create, search, pull, publish, review, and continue versioned context from terminal workflows and unattended agents. The package is dependency-free, MIT licensed, and requires Node.js 22.13 or newer.

• npm: https://www.npmjs.com/package/handover-sh
• Setup guide: https://handover.sh/install?utm_source=github&utm_medium=referral&utm_campaign=cli_v0_1_0
• CLI documentation: https://handover.sh/docs/cli-and-api?utm_source=github&utm_medium=referral&utm_campaign=cli_v0_1_0
• Security model: https://handover.sh/security?utm_source=github&utm_medium=referral&utm_campaign=cli_v0_1_0

Verify the active identity with handover whoami before reading or publishing company context.

Handover Agent Skills v1.1.0

Verify before protected work

This release expands the open Handover collection to five skills:

• handover-verify checks the connected MCP or CLI identity, boundary, workspace, role, scopes, and one protected read before an agent writes.
• handover-publish captures durable context.
• handover-resume continues another actor's work.
• handover-review handles revision-anchored feedback.
• handover-governance scopes and audits agent access.

The verification workflow is read-only by default, never exposes credentials, classifies the first failed layer, and includes an optional two-identity continuity proof.

Catalog: https://skills.handover.sh/?utm_source=github&utm_medium=referral&utm_campaign=agent_skills_launch

Verification skill: https://skills.handover.sh/handover-verify?utm_source=github&utm_medium=referral&utm_campaign=handover_verify_skill

Install: npx skills add 44-pixels/handover-mcp

Handover Agent Skills v1.0.0

Four open, installable workflows for continuing work across people and AI agents.

Included skills:

• handover-publish: preserve decisions, files, evidence, and the next action
• handover-resume: retrieve the canonical revision and continue without replaying a chat
• handover-review: create and resolve revision-anchored feedback
• handover-governance: scope, test, and revoke human and service-agent access

Install the collection:

npx skills add 44-pixels/handover-mcp

Every skill is MIT licensed and source-readable. Skills provide the reusable workflow; Handover MCP or the CLI provides authenticated capabilities and attributable history.

• Workflow catalog: https://skills.handover.sh/?utm_source=github&utm_medium=referral&utm_campaign=agent_skills_launch
• Public source: https://github.com/44-pixels/handover-mcp/tree/main/skills
• Agent Skills and MCP guide: https://handover.sh/guides/agent-skills-and-mcp?utm_source=github&utm_medium=referral&utm_campaign=agent_skills_mcp
• Connect Handover: https://handover.sh/install?utm_source=github&utm_medium=referral&utm_campaign=agent_skills_launch
• Official MCP Registry entry: https://registry.modelcontextprotocol.io/v0.1/servers?search=sh.handover%2Fhandover