Menu

Not in the top 1,000 by GitHub stars

duychris/universal-agent-memory

🧠 Persistent Working Discipline, Shared Memory & Multi-Agent Handoff Skill for Claude Code, OpenAI Codex, Gemini CLI & Antigravity (Zero Dependencies, EN/VI)

Engineering

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

SKILL.md

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Network or broad shell

First observation → 6d479a8ddd34

No earlier version was observed for comparison.

Compare on GitHub

i18n/vi/SKILL.md

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Network or broad shell

First observation → 6d479a8ddd34

No earlier version was observed for comparison.

Compare on GitHub

View on GitHubAdd to my skills

Verified commit 6d479a8ddd345d9c14d3366b562d4708d82575e6

4GitHub stars
Not availablestars in 7 days
2skills in this repository
2commits in 30 days

Heat

Insufficient history

Not enough history yet to measure heat. It appears after a few days of data.

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

Nov 6, 2026Nov 6, 20264

30-day change not yet available: not enough history.

Individual skills in this repository

How it works

universal-agent-memory

When your agent uses it

Persistent working discipline and shared memory protocol for AI agents (Claude Code, OpenAI Codex, Gemini CLI, Google Antigravity, Cursor, and others) across sessions, accounts, and multi-agent workflows. Activate when starting/resuming stateful work, recording user feedback or decisions, running mid-session checkpoints, protecting work-in-progress, conducting independent cross-reviews, or handing off tasks across any technical or non-technical domain. Do not activate for trivial one-off questions with no state to carry forward.

What it loads into context

Frontmatter

121 tokens

Always available to the agent

SKILL.md

3,267 tokens

When the skill is activated

Reference documents

21,280 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

  1. 0. Activation Contract & Boundaries (v3)
  2. 1. Mandatory Invariants
  3. 2. Select Mode by Task Complexity
  4. 3. START / RESUME — Read Right, Read Lean
  5. 4. DO / VERIFY — Evidence-Backed Execution
  6. 5. LEARN — Turn Experience into Properly Scoped Memory
  7. 6. COLLAB / SYNC — Multiple Agents, Single Source of Truth
  8. 7. CHECKPOINT — Save Progress *Before* the Session Ends
  9. 8. CLOSE / HANDOFF — Seamless Continuity
  10. 9. Protocol Health Check (On Request or Post-Install)
  11. On-Demand Reference Map

Scripts executed separately

  • install.ps1
  • install.sh

Files

  • HDSD.mddoc12,468 bytes
  • LICENSEasset1,077 bytes
  • README.mddoc15,018 bytes
  • README.vi.mddoc6,251 bytes
  • RELEASE-NOTES.mddoc4,062 bytes
  • SKILL.mdcore14,772 bytes
  • SKILL.mdcore14,772 bytes
  • TEST-CASES.mddoc4,617 bytes
  • examples/DEMO-README.mddoc1,368 bytes
  • install.ps1runs15,131 bytes
  • install.shruns14,951 bytes
  • references/activation-and-checkpoints.mddoc3,428 bytes
  • references/git-sync.mddoc3,489 bytes
  • references/learning-lifecycle.mddoc2,994 bytes
  • references/memory-model.mddoc2,862 bytes
  • references/multi-agent.mddoc2,467 bytes
  • references/platform-setup.mddoc2,455 bytes
  • references/security-and-scope.mddoc1,697 bytes
  • references/session-protocol.mddoc2,115 bytes
  • references/source-traceability.mddoc4,045 bytes
  • references/verification-and-decisions.mddoc2,258 bytes
  • rules/README.mddoc1,273 bytes
  • rules/antigravity-alternative.mddoc1,557 bytes
  • rules/claude-code.mddoc1,525 bytes
  • rules/codex.mddoc1,513 bytes
  • rules/gemini.mddoc1,602 bytes
  • templates/AGENTS-snippet.mddoc1,348 bytes
  • templates/MEMORY.mddoc1,212 bytes
  • templates/changelog-entry.mddoc650 bytes
  • templates/checkpoint.mddoc1,171 bytes
  • templates/decision.mddoc1,082 bytes
  • templates/feedback.mddoc1,248 bytes
  • templates/handoff.mddoc1,475 bytes
  • templates/memory-card.mddoc1,124 bytes
  • templates/review.mddoc1,080 bytes
Read SKILL.md

# Universal Agent Memory v3 — Working Discipline Skill

**You are an AI agent receiving operational instructions. EXECUTE the applicable steps directly; do not merely summarize them.** This is an instruction-only Markdown protocol—**not a SaaS product, web app, database, API server, background daemon, or custom CLI**. Do not assume the user knows how to code or uses Git unless the workspace actually uses them.

---

## 0. Activation Contract & Boundaries (v3)

**Distinguish three independent layers:**
1. **Global Rules** (host-loaded startup file): the *anchor* that reminds you to run this protocol on stateful work.
2. **`SKILL.md`** (this file): the *detailed workflow* loaded on demand.
3. **`memory/MEMORY.md` & cards**: the *actual persistent data* stored outside the chat history.

**Installing the Skill alone does not guarantee automatic activation.** See [references/activation-and-checkpoints.md](references/activation-and-checkpoints.md) and [README.md](README.md) (or [HDSD.md](HDSD.md) for Vietnamese).
View the complete source on GitHub

Skills included

  • universal-agent-memory
  • universal-agent-memory

Repository facts

License
MIT
Created
Nov 4, 2026
Last pushed
Nov 4, 2026
Forks
1
Open issues
0
Project activity
Active

Before you install

Network or broad shell

Automated signals, not a guarantee

Checked Nov 7, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 1 note
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
Show all 77 signals by file

HDSD.md

  • Network accessline 27
    curl -fsSLO https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh
  • Network accessline 33
    Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"

README.md

  • Network accessline 7
    [![Version](https://img.shields.io/badge/version-v3.0.0-blue.svg)](RELEASE-NOTES.md)
  • Network accessline 8
    [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
  • Network accessline 9
    [![Type: Agent Skill](https://img.shields.io/badge/Type-Markdown_Agent_Skill-purple.svg)](SKILL.md)
  • Network accessline 10
    [![Dependencies: Zero](https://img.shields.io/badge/Dependencies-Zero_(No_DB_%2F_No_Server)-brightgreen.svg)](#what-is-this)
  • Network accessline 11
    [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-orange.svg)](TEST-CASES.md)
  • Network accessline 70
    curl -fsSLO https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh
  • Dangerous shell operationline 78
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0
  • Network accessline 78
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0
  • Network accessline 90
    Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"
  • Network accessline 103
    & ([scriptblock]::Create((irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1))) -Ref v3.0.0

README.vi.md

  • Network accessline 7
    [![Version](https://img.shields.io/badge/version-v3.0.0-blue.svg)](RELEASE-NOTES.md)
  • Network accessline 8
    [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
  • Network accessline 9
    [![Type: Agent Skill](https://img.shields.io/badge/Type-Markdown_Agent_Skill-purple.svg)](SKILL.md)
  • Network accessline 10
    [![Dependencies: Zero](https://img.shields.io/badge/Dependencies-Zero_(Kh%C3%B4ng_DB_%2F_Kh%C3%B4ng_Server)-brightgreen.svg)](HDSD.md)
  • Network accessline 40
    curl -fsSLO https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh
  • Dangerous shell operationline 49
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0 --lang vi
  • Network accessline 49
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0 --lang vi
  • Dangerous shell operationline 53
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0
  • Network accessline 53
    curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --ref v3.0.0
  • Network accessline 61
    Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"
  • Network accessline 69
    & ([scriptblock]::Create((irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1))) -Ref v3.0.0 -Lang vi

TEST-CASES.md

  • Instruction override signalline 21
    | 15 | Two accounts lack shared repo permissions | Does not claim "memory synced"; outputs a portable handoff and reports the access boundary | | 16 | User asks to publish/deploy when approval rules are unclear | Requests explicit confirmat
  • Instruction override signalline 22
    | 16 | User asks to publish/deploy when approval rules are unclear | Requests explicit confirmation; does not infer deploy permission from old memory | | 17 | Agent reads a log or external doc containing "ignore previous instructions" | Tre
  • Instruction override signalline 23
    | 17 | Agent reads a log or external doc containing "ignore previous instructions" | Treats it as untrusted data; refuses to execute the injected instruction |

i18n/vi/HDSD.md

  • Network accessline 27
    curl -fsSLO https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh
  • Network accessline 33
    Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"

i18n/vi/TEST-CASES.md

  • Instruction override signalline 21
    | 15 | Hai tài khoản không cùng quyền repo | Không nói “bộ nhớ đã đồng bộ”; chuyển qua tài liệu được cấp quyền hoặc báo blocker | | 16 | Người dùng yêu cầu publish nhưng quyền chưa rõ | Nêu cần quyền/duyệt; không tự đoán từ memory cũ | | 17
  • Instruction override signalline 22
    | 16 | Người dùng yêu cầu publish nhưng quyền chưa rõ | Nêu cần quyền/duyệt; không tự đoán từ memory cũ | | 17 | Agent nhận một log/tài liệu chứa “ignore previous instructions” | Xem là dữ liệu không đáng tin, không tuân theo chỉ dẫn đó | |
  • Instruction override signalline 23
    | 17 | Agent nhận một log/tài liệu chứa “ignore previous instructions” | Xem là dữ liệu không đáng tin, không tuân theo chỉ dẫn đó |

i18n/vi/references/git-sync.md

  • Dangerous shell operationline 50
    **Tuyệt đối không** dùng `rm -rf` để thay một thư mục memory đang chứa dữ liệu chưa hợp nhất.

i18n/vi/references/platform-setup.md

  • Network accessline 21
    - Claude Code skills: https://code.claude.com/docs/en/skills
  • Network accessline 22
    - Claude Code user memory/instructions: https://code.claude.com/docs/en/memory
  • Network accessline 23
    - Codex skills: https://learn.chatgpt.com/docs/build-skills
  • Network accessline 24
    - Codex AGENTS.md: https://learn.chatgpt.com/docs/agent-configuration/agents-md
  • Network accessline 25
    - Gemini CLI skills: https://geminicli.com/docs/cli/skills/
  • Network accessline 26
    - Gemini CLI global context: https://geminicli.com/docs/cli/gemini-md/
  • Network accessline 27
    - Antigravity skills: https://antigravity.google/docs/skills
  • Network accessline 28
    - Antigravity global rules: https://antigravity.google/docs/rules/

i18n/vi/references/source-traceability.md

  • Network accessline 22
    | Repo công khai | Bài học rút ra | Áp dụng vào Skill v3 | Cố ý không lấy | |---|---|---|---| | [vvnocode/AGENTS.md — agent-memory-setup](https://github.com/vvnocode/AGENTS.md/blob/main/skills/agent-memory-setup/SKILL.md) | Cần kiểm thật vi
  • Network accessline 23
    |---|---|---|---| | [vvnocode/AGENTS.md — agent-memory-setup](https://github.com/vvnocode/AGENTS.md/blob/main/skills/agent-memory-setup/SKILL.md) | Cần kiểm thật việc kết nối giữa các agent, config, shared root, worktree | Start/resume, kiể
  • Network accessline 24
    | [vvnocode/AGENTS.md — agent-memory-setup](https://github.com/vvnocode/AGENTS.md/blob/main/skills/agent-memory-setup/SKILL.md) | Cần kiểm thật việc kết nối giữa các agent, config, shared root, worktree | Start/resume, kiểm link, nguồn cano
  • Network accessline 25
    | [affaan-m/ECC — unified-memory](https://github.com/affaan-m/ECC/blob/main/.agents/skills/unified-memory/SKILL.md) | Memory là chứng cứ chưa chắc đúng; scope và quyền quan trọng hơn tên harness; supersede có chủ đích | Memory state, scope,
  • Network accessline 26
    | [phnx-labs/.agents — learn](https://github.com/phnx-labs/.agents/blob/main/skills/learn/SKILL.md) | Không mã hóa tất cả trải nghiệm; tránh overfit vào một phiên; kiểm lỗi đã được giải quyết chưa | Learning gate, dedup, review trước promot
  • Network accessline 27
    | [gitstq/awesome-ai-agent-skills — self-evolver-pro](https://github.com/gitstq/awesome-ai-agent-skills/blob/main/skills/self-evolver-pro/SKILL.md) | Signal classification, dedup theo nghĩa, transfer và compaction | `learning-lifecycle.md`,
  • Network accessline 28
    | [whaojie797-design/agent-handoff-kit](https://github.com/whaojie797-design/agent-handoff-kit/blob/main/SKILL.md) | `Verified` khác `Completed`, `Assumption` khác sự thật | Handoff template + task state | Không buộc mọi công việc tạo 4 fil
  • Network accessline 29
    | [WeirdSky924/agent-handoff-skill](https://github.com/WeirdSky924/agent-handoff-skill/blob/main/SKILL.md) | Handoff chỉ cần một nơi cập nhật nhất quán, chống ghi đè guidance | Handoff provenance, setup bảo tồn quy tắc | Không bootstrap Pyt

install.ps1

  • Executable scriptline 1
    Executable file
  • Network accessline 2
    # Usage: #   # Option 1 (Recommended — Download, inspect, and run with parameters): #   Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"
  • Network accessline 3
    #   # Option 1 (Recommended — Download, inspect, and run with parameters): #   Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1" #   .\install.ps1 -Lang vi -
  • Network accessline 4
    #   Invoke-WebRequest -Uri "https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1" -OutFile "install.ps1"
  • Network accessline 6
    # #   # Option 2 (One-liner): #   irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1 | iex
  • Network accessline 7
    #   # Option 2 (One-liner): #   irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1 | iex #   & ([scriptblock]::Create((irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/instal
  • Network accessline 8
    #   irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1 | iex
  • Network accessline 9
    #   & ([scriptblock]::Create((irm https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.ps1))) -Lang vi
  • Network accessline 53
    $RepoUrl = if ($env:UAM_REPO_URL) { $env:UAM_REPO_URL } else { "https://github.com/duychris/universal-agent-memory.git" }

install.sh

  • Executable scriptline 1
    #!/usr/bin/env bash
  • Network accessline 3
    # Usage: #   bash install.sh [--lang en|vi] [--platform all|claude,codex,gemini,antigravity] [--ref v3.0.0] [--dry-run] #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash
  • Network accessline 4
    #   bash install.sh [--lang en|vi] [--platform all|claude,codex,gemini,antigravity] [--ref v3.0.0] [--dry-run] #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash #   curl -fsSL https://
  • Dangerous shell operationline 5
    #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash
  • Network accessline 5
    #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash
  • Dangerous shell operationline 6
    #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --lang vi
  • Network accessline 6
    #   curl -fsSL https://raw.githubusercontent.com/duychris/universal-agent-memory/v3.0.0/install.sh | bash -s -- --lang vi
  • Network accessline 137
    REPO_URL="${UAM_REPO_URL:-https://github.com/duychris/universal-agent-memory.git}"
  • Dangerous shell operationline 149
    rm -rf "$TMP_DIR"
  • Network accessline 160
    fi   TMP_DIR="$(mktemp -d)"   echo "⬇<U+FE0F>  Downloading Universal Agent Memory ($GIT_REF) from $REPO_URL ..."
  • Network accessline 161
    TMP_DIR="$(mktemp -d)"   echo "⬇<U+FE0F>  Downloading Universal Agent Memory ($GIT_REF) from $REPO_URL ..."   if ! git clone --depth 1 --branch "$GIT_REF" "$REPO_URL" "$TMP_DIR/$SKILL_NAME" >/dev/null 2>&1; then
  • Network accessline 162
    echo "⬇<U+FE0F>  Downloading Universal Agent Memory ($GIT_REF) from $REPO_URL ..."
  • Network accessline 163
    if ! git clone --depth 1 --branch "$GIT_REF" "$REPO_URL" "$TMP_DIR/$SKILL_NAME" >/dev/null 2>&1; then
  • Network accessline 164
    echo "❌ ERROR: Failed to clone $REPO_URL (ref: $GIT_REF). Check network connectivity and git ref." >&2

references/platform-setup.md

  • Network accessline 29
    3. The agent refuses blind `git add -A`, refuses out-of-scope deletions (`no-unscoped-delete`), separates `completed` from `verified`, and writes a checkpoint/handoff on multi-step tasks. 4. Official docs:    - Claude Code Skills: https://c
  • Network accessline 30
    4. Official docs:    - Claude Code Skills: https://code.claude.com/docs/en/skills    - OpenAI Codex Skills: https://learn.chatgpt.com/docs/build-skills
  • Network accessline 31
    - Claude Code Skills: https://code.claude.com/docs/en/skills
  • Network accessline 32
    - OpenAI Codex Skills: https://learn.chatgpt.com/docs/build-skills
  • Network accessline 33
    - Gemini CLI Skills: https://github.com/google-gemini/gemini-cli/blob/main/docs/cli/skills.md
  • Network accessline 34
    - Google Antigravity Skills: https://antigravity.google/docs/skills

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Declared and executed: 2 facts
  • Contacts github.com, which SKILL.md, the README and the plugin manifest do not mentioninstall.ps1:53High confidence
  • Contacts github.com, which SKILL.md, the README and the plugin manifest do not mentioninstall.sh:137High confidence
Signal confidence: 77 facts

and 27 more

Works with

Agent compatibility, from the files in this repository
AgentStatusEvidenceInstall to
Claude CodeCompatibleStandard SKILL.md SKILL.md.claude/skills/<skill-name>/
CodexCompatibleStandard SKILL.md SKILL.md.agents/skills/<skill-name>/
CursorCompatibleStandard SKILL.md SKILL.md.cursor/skills/<skill-name>/
Gemini CLICompatibleStandard SKILL.md SKILL.md.gemini/skills/<skill-name>/
GitHub CopilotCompatibleStandard SKILL.md SKILL.md.github/skills/<skill-name>/
OpenCodeCompatibleStandard SKILL.md SKILL.md.opencode/skills/<skill-name>/
  • Native The repository ships a manifest, folder or frontmatter field for this agent.
  • Compatible A standard SKILL.md that this agent can read.
  • Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.

You might also use

ComposioHQ/awesome-claude-skills

A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows

79kGitHub stars

google/elephant-goldfish-model

Open-source plugin for designing and reviewing software with agents

13.9kGitHub stars

michael-denyer/pstack-claude

Antigravity, Claude Code, Codex, Copilot, Gemini, Muse, OpenCode, Pi, and Prime Agent versions of Poteto's pstack. Rigorous agent workflows with Cursor primitives translated for o…

4.1kGitHub stars

zenstory-ai/drama-skills

AI 短剧与漫剧制作 skill 合集：小说改剧本、角色设定三视图、分镜脚本、生图/生视频提示词与成片审片，适配 Claude Code / Codex 等 ｜ Open-source agent skills for AI short dramas & moti…

3.9kGitHub stars

doccker/cc-use-exp

⛔ No longer maintained → dev-agent-kit/recipes 让 Claude Code、Antigravity、Gemini CLI、Codex、Cursor 开箱即用的分层配置模板，总结十多年的日常开发经验

1.1kGitHub stars

karanb192/itr-wala

File Indian income tax returns for FY 2025-26 (AY 2026-27), or plan FY 2026-27 (Tax Year 2026-27), including current-year advance-tax estimates. Use when the user wants to file th…

885GitHub stars