| name | push |
| description | Operate a Push personal assistant, inspect its health and jobs, and author or validate Push job runbooks. Use when a request concerns Push setup, CLI commands, assistant files, scheduled jobs, or delivery behavior. |
| license | MIT |
| compatibility | Requires the Push CLI and an initialized assistant repository. |
| metadata | {"push-managed-version":"3"} |
Push
Use Push as the gateway around the configured Claude Code, Codex, or Pi
runtime. Push owns channels, scheduling, history, security checks, and delivery.
The agent runtime owns reasoning, tools, skill execution, permissions, MCP
servers, and authentication.
Work in the assistant repository
- Treat
SOUL.md as user-owned identity and AGENTS.md as user-owned repository
guidance. Do not edit either unless the user asks.
- Read
context/README.md first when durable user context is relevant. Keep
useful durable notes under context/ and evaluation criteria under evals/.
- Store complete job runbooks under
jobs/.
- Treat
skills/push/ and the push links under .agents/skills/ and
.claude/skills/ as Push-managed. Do not edit them. Codex and Pi share the
.agents/skills/ discovery path.
- Keep credentials, configuration containing credentials, sessions, databases,
audit logs, and job runtime files outside the assistant repository.
Use the current CLI
Run push help when the accepted command forms are unclear. The stable commands
are:
push help
push version
push init [path]
push
push doctor
push status
push paths
push reload or push restart
push job validate
push job list
push job show <name>
push job run <name>
push job runs [<name>]
push job reviews [<name>]
All commands accept --config <path>, although service status and restart use
the installed service definition. Inspection commands support --json where
documented by push help: help, version, doctor, status, paths, job validation,
job inspection, runs, and reviews. Commands that start the gateway, change
service state, scaffold files, or run a job reject JSON mode. The first upgraded
inspection that loads configuration and job state records the one-time schedule
migration baseline; follow the release upgrade notes before running it. Prefer
JSON for automation, but still treat paths, handles, and operational metadata
as sensitive. Never expose tokens, message content, or sensitive runtime state
in diagnostics or replies.
Author jobs safely
- Read the job format and existing runbooks before editing.
- Write the complete Markdown runbook directly under
jobs/ when the user
asks to create or change a job.
- Keep secrets out of the runbook. Use the configured service or agent
environment for credentials.
- Run
push job validate after every job change. Do not claim success if
validation fails.
- Saving a new or changed enabled schedule does not activate it. Tell the user
that Push will present the exact revision for separate owner review.
- Use
push job show <name> to inspect the installed result,
push job reviews <name> to inspect schedule activation state, and
push job runs <name> to inspect execution and delivery history.
- Run
push job run <name> only when the user asked for the job to execute or
when execution is a clearly authorized part of the task.
Reply normally
For an ordinary conversation, return the final reply normally. Push sends it
back through the originating channel. Do not invoke the Push CLI merely to send
a chat reply.