Skip to content
Skills Library
AI & agents · Tool integration

send

/send

Use when a user wants to start a new Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok session from a coSlash session.

Repository
coslash ↗
Licence
MIT
Last repository change
27.10.2026

Assessment and security

91/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content26 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when a user wants to start a new Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok session from a coSlash session.

Original author description

Use when a user wants to start a new Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok session from a coSlash session.

How it works

SKILL.md has no level-two headings · 28 lines, 721 words

SKILL.md preview
---
name: send
description: Use when a user wants to start a new Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok session from a coSlash session.
---

# coSlash send

If the user means the current session, resolve it from the host:

| Host | Selector |
| --- | --- |
| Claude Code | `claude:$CLAUDE_CODE_SESSION_ID` when that variable is set. If it is unset, list sessions for Claude. |
| Codex | `codex:$CODEX_SESSION_ID` when that variable is set. If it is unset, list sessions for Codex. |
| OpenCode | `coslash sessions --agent opencode --recent 20 --json` |
| Cursor CLI | `coslash sessions --agent cursor --recent 20 --json` |
| Pi | `coslash sessions --agent pi --recent 20 --json` |
| Grok | `coslash sessions --agent grok --recent 20 --json` |

Use a returned `selector` only when it uniquely identifies the intended session; if multiple sessions could match, ask the user to choose. Do not assume the newest session is current. For a named session, use the `selector` returned by `coslash sessions [query] --json`. Then run `coslash send <agent>:<session> --to claude|codex|opencode|cursor|pi|grok [message]` with the selected `selector`, the user's target, and optional message. The message is the task that the user gives to the target agent, for example "do an analysis". Pass the message as one quoted argument. For Cursor, the CLI copies the handoff and task to the clipboard and opens Cursor CLI; tell the user to paste them there.

Use the `coslash` executable found on `PATH` and preserve the inherited `COSLASH_HOME`. Do not substitute a repository-relative binary or `go run` unless the user explicitly asks to test a source build.

If a command reports `Error: usage:`, compare the usage it prints with the command you ran. If that usage does not list a subcommand or flag that you used, the installed coSlash is older than this skill. Run `coslash --version`, tell the user to update coSlash the same way they installed it, for example `brew upgrade coslash`, and stop. Otherwise, the arguments are not valid: return stderr and the exit status. Do not retry with other flags.

In Codex, if the command reports `coSlash app is not running`, `coSlash app is not reachable`, or `operation not permitted` for coSlash storage, request escalated execution and retry the exact command outside the sandbox. Use the same elevated execution for any startup and retries below. If permission is denied, stop and report that coSlash needs access to its local files and loopback address.

If the command reports `coSlash app is not running` after the Codex retry, when applicable, resolve the `coslash` executable from `PATH`, launch that executable directly as `coslash --no-open --port 0` in the background with output redirected to temporary logs created with `mktemp -t coslash.XXXXXX` on macOS/Linux or `New-TemporaryFile` on Windows PowerShell, and retain its PID. On Windows PowerShell, use `Start-Process` with `-WindowStyle Hidden`, `-PassThru`, and separate stdout/stderr redirection; retain the returned process object and include both logs in startup failures. Do not launch for other errors. Every two seconds, retry the exact original command while it reports `coSlash app is not running`, until three minutes pass. The command reports that error before it contacts the app, so a retry has no side effect. If the retained process exits because another coSlash app is already running, keep retrying for that app; for any other early exit, return the startup logs.

When the command reports anything else, stop retrying and use that result. Keep the temporary logs while the retained process runs, and arrange their removal after that process exits; if it already exited as a startup contender, remove the logs when the retries end. On startup failure or timeout, stop the retained process if it is still running (SIGTERM on macOS/Linux, or `Stop-Process -InputObject $process` followed by `$process.WaitForExit()` on Windows PowerShell, using the object returned by `Start-Process`), return the startup logs, then remove the logs. Otherwise leave the instance running by default. If the user asks to close it after the command, arrange cleanup before launching so interruption also stops that same process object using the same platform command, then wait for it to exit and remove the logs. Report cleanup failures on stderr without changing command stdout. Never stop an instance that was already running or a process found by name, port, or a PID file.

Return stdout unchanged. If the command fails, return stderr and the exit status, then suggest `coslash doctor --json`. Do not construct the handoff or launch the agent yourself.

Versions

RepositoryScoreLicence
This page · centauri-ai/coslash91MIT
xhluca/agent-talk100MIT
synaptiai/agent-capability-standard100Apache-2.0
SayandipBagchi/release-notes-plugin90MIT

Database snapshot: 13.11.2026 · package version e00b8801cb1d (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 91review
    Use when a user wants Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok to review a local coSlash session.
  • 100ask-other-agent
    Gets an opinion from, or hands a task to, Codex, Claude Code, Grok, or Cursor. Use when the user asks to ask one of them.
  • 91handoff
    Use when a user wants the canonical coSlash handoff or context for a specific session.
  • 95agents-cli
    Run and manage many AI coding agents (Claude Code, Codex, Cursor, Gemini/Antigravity, Grok, Kimi, OpenCode, Droid, …) from one CLI — the `agents` command. Use this to run multiple coding agents in parallel, manage multiple Claude Code accounts, keep working when I hit my usage…
  • 100harness-master
    Audit harness configs, discover gaps, usage signals, apply approved fixes. Use when tuning Claude, Codex, Copilot, Cursor, Gemini, Grok, OpenCode, Cherry, or LM Studio. NOT agents, MCP, or app telemetry.
  • 90t3-delegate
    Delegates a task to another provider or model as a T3 Code child task through `delegate_task`. Use inside T3 Code whenever a skill would otherwise shell out to `codex`, `claude -p`, `grok -p`, or `cursor-agent`.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.