# Cmd Latest Msg

> Store or retrieve the latest agent message to /tmp/agents/{agent}/

- **Type:** Skill
- **Install:** `agentstack add skill-olshansk-agent-skills-cmd-latest-msg`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [Olshansk](https://agentstack.voostack.com/s/olshansk)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [Olshansk](https://github.com/Olshansk)
- **Source:** https://github.com/Olshansk/agent-skills/tree/main/skills/cmd-latest-msg

## Install

```sh
agentstack add skill-olshansk-agent-skills-cmd-latest-msg
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Latest Message 

Persist the last agent message to disk for cross-agent visibility.

## Instructions

1. Parse the user's input after `/cmd-latest-msg`.

2. **`help`** (or no argument) — print usage and stop:

   ```
   Usage:
     /cmd-latest-msg save             — save your last message
     /cmd-latest-msg use              — read latest from the current agent
     /cmd-latest-msg use       — read latest from claude|codex|gemini
     /cmd-latest-msg help             — show this usage
   ```

3. **`save`** — save the last assistant message:

   a. Determine which agent you are from runtime context:
      - Claude Code → `claude`
      - Codex CLI → `codex`
      - Gemini CLI → `gemini`
      - If inference fails, default to `codex`

   b. Create the output directory:

   ```bash
   mkdir -p /tmp/agents/
   ```

   c. Capture your **last assistant message** — the response you gave immediately before the user invoked this skill.

   d. Write it to two files using the Write tool:
      - `/tmp/agents//latest.md` — always overwrite
      - `/tmp/agents//.md` — timestamped copy (use `date +%s` to get the timestamp)

   e. Confirm with:

   ```
   Stored to:
     /tmp/agents//latest.md
     /tmp/agents//.md
   ```

4. **`use`** — display the latest message for the current agent:

   a. Infer the current agent using the same runtime-context rules as `save`.

   b. Read and display `/tmp/agents//latest.md`.

   c. If the file doesn't exist, say: `No messages found for .`

5. **`use `** — display another agent's latest message:

   a. Validate `` is one of: `claude`, `codex`, `gemini`.

   b. Read and display `/tmp/agents//latest.md`.

   c. If the file doesn't exist, say: `No messages found for .`

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [Olshansk](https://github.com/Olshansk)
- **Source:** [Olshansk/agent-skills](https://github.com/Olshansk/agent-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-olshansk-agent-skills-cmd-latest-msg
- Seller: https://agentstack.voostack.com/s/olshansk
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.