- Home
- Skills
- Agents & Orchestration
- ShareMemory
More screenshots
Works with the AI Agents you already use
ShareMemory
Shared project memory and handoffs for Claude Code and Codex, stored in your repository.
$8
ShareMemory
Example session with this skill installed
Initialize ShareMemory in a fresh local project. Use English memory and leave Git recovery disabled. Then run memory status and verify that a second process can read a handoff from the same AI_MEMORY/ folder.
Verification example below: a deterministic local fixture using the packaged templates and helpers; it is not a transcript from two live AI models.
- Read your context and instructions
- Compiled the sharememory
Verified local package example
The packaged v1.3 files were extracted into a disposable project and exercised with two fixture identities. This is a local file/lock test, not a live-agent transcript.
Initialized project structure
target-project/
├── AGENTS.md
├── CLAUDE.md
├── MEMORY_PROTOCOL.md
├── scripts/
│ ├── check_memory.sh
│ └── memory_lock.sh
└── AI_MEMORY/
├── CONFIG.md
├── PROJECT.md
├── DECISIONS.md
├── TASKS.md
├── LEARNINGS.md
├── SYNC_LOG.md
└── archive/
Complete verification summary
| Check | Result |
|---|---|
| Existing template/lock acceptance suite | 21/21 passed |
| ZIP and extracted-fixture checks | 20/20 passed |
| Both agent skill paths | Resolved to the same extracted SKILL.md |
| Initialization | Passed memory lint |
| First fixture identity holds the lock | Second identity is blocked |
| Wrong-identity lock release | Rejected |
| First identity writes a handoff | Independent process reads the same note |
| Second identity acknowledges the handoff | Succeeds after the first releases the lock |
| Updated log versus prior boot receipt | Status reports the change |
| Final state | Memory lint passed; no held lock or Git repository |
Observed helper output included
✓ lock acquired by FixtureA
✓ memory OK
✓ lock released
✓ boot recorded for FixtureB
✓ lock acquired by FixtureB
✓ lock released
Use it in your project
After installing the complete skill folder, ask "init memory", choose your language and Git preference, then use "memory status" and "update memory" during coding handoffs. Both coding agents must open the same project directory.
Connects securely to your tools. The creator never sees your data.
About this skill
ShareMemory gives Claude Code and Codex one shared, project-local memory folder. Keep decisions, active tasks and handoff notes in plain Markdown so the next coding session can pick up the work.
What you get
- One AI_MEMORY/ folder per project with project context, decisions, tasks, learnings and a daily handoff log.
- Shared startup instructions in AGENTS.md and CLAUDE.md, plus a MEMORY_PROTOCOL.md that describes how to read and update memory.
- Two Bash helpers: an advisory write lock with stale-lock recovery, and a linter for memory structure, size limits and basic secret-pattern checks.
- Instructions for initializing, updating, checking, consolidating, repairing and migrating project memory.
Getting started
- Install the complete skill folder in the project where you want shared memory. For Claude Code and Codex together, use one canonical skill copy and the documented symlink layout.
- Ask your coding agent to "init memory". Choose the memory language and whether to enable Git history recovery.
- Open the same project directory with the other agent. Ask for "memory status" to inspect the shared state.
- Use "update memory" to record a handoff or refresh the current project context.
Designed for local coding workflows
Both agents need access to the same project files. The workflow requires local file read/write access, Bash and standard shell utilities. It does not require a hosted memory service, database or API keys.
The lock is advisory: it helps agents that follow the protocol coordinate their writes. It cannot stop an agent that ignores the rules or make simultaneous editing universally safe. ShareMemory is project-scoped and does not provide cross-project memory or built-in remote synchronization.
Included buyer guide
The package includes BUYER_GUIDE.md with setup, daily usage, troubleshooting and license information. Detailed operation steps are in references/operations.md, supplied with the skill.
Verification
The v1.3 package passed 21 existing acceptance checks and 20 package/local-fixture checks. The fixture covered initialization, shared file paths, handoffs and lock behavior. These checks did not run two live AI agents or validate Agensi connector delivery.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scanned
Verified clean 2 days ago
- Passed all security checks, Safe to install
Requires