Postnavo CLI + agent skill

Why a command line when there is an MCP server

A coding agent already works in a shell, so a command is the shortest path for it. The same command runs where there is no agent at all: a script, a cron job, a CI pipeline. Postnavo is launching with a CLI for those cases and an MCP server for the chat apps.

What an agent skill is

The Agent Skills specification defines a skill as “a directory containing, at minimum, a SKILL.md file”. The file opens with YAML frontmatter: a name, and a description of what the skill does and when to use it.[1]

Agents load the name and description of every skill at startup and read the rest only when a skill is activated. The specification calls that progressive disclosure.[1]

Postnavo is launching with an installable skill for its CLI: a small instruction package that tells a coding agent which command to run, and when. Where it installs for each agent is published at launch.

What an agent asks before it runs a command

Claude Code asks before it runs a shell command, apart from a built-in set of read-only commands, and can save your answer as a rule for the repository.[6]

Codex runs with network access turned off by default and asks for approval before a command that needs the network. A command that hands a post to a scheduler does.[7]

OpenClaw’s docs tell you to treat third-party skills as untrusted code and to read them before enabling. Gemini CLI shows a confirmation prompt when it activates a skill.[3][5]

Postnavo is launching with an approval setting on its own side as well, so a post can wait for a person whichever agent ran the command.

In CI: a token, a secret and a schedule

In a pipeline nobody can sign in by hand. Postnavo is launching with API tokens for that case, used in place of the interactive login.

On GitHub Actions a token belongs in a secret. You create one in the repository’s Settings, under “Secrets and variables”, Actions, with “New repository secret”.[8]

A workflow can run on a schedule written in POSIX cron syntax. GitHub warns that the schedule event “can be delayed during periods of high loads”, names the start of every hour as such a time, and gives once every 5 minutes as the shortest interval.[9]

A late runner matters less when the post carries its own time. The workflow hands the post over together with its publish time, and the calendar Postnavo is launching with does the waiting.