generate-content

Generate Content Skill

Produce platform-specific draft text using PostAll's generator.

Inputs

The planner provides these args (read from args):

• topic (required) — what the content is about, in plain language.
• platforms (required) — comma-separated list, e.g. twitter,linkedin,wechat.
• style (optional) — professional / casual / technical / playful. Default: professional.
• url (optional) — if the wish included a URL, pass it so postall extracts content from it.

If topic or platforms is missing, return an error final message and do not call postall-agent.

Execution

Invoke postall-agent via run_bash. Build the command from the args:

postall-agent generate \
  --topic "<TOPIC>" \
  --platforms "<PLATFORMS>" \
  [--style "<STYLE>"] \
  [--url "<URL>"]

The CLI prints a JSON object to stdout. Capture it.

Output handling

Parse the JSON. Key fields you must return to the user:

• draft_id — needed by downstream skills (image, publish).
• platforms — confirmation of which platforms got drafts.
• content_preview — first 200 chars of each platform's draft for the user to skim.

Return a single concise final message in this shape:

Draft created.

draft_id: <ID>
platforms: <PLATFORMS>

Twitter:
  <first ~200 chars>

LinkedIn:
  <first ~200 chars>

If postall-agent returns success: false, surface the error message to the user verbatim and do NOT pretend the draft was created.

Composability notes

Other skills in this pack consume draft_id. The wish-runner threads it through the args of the next skill automatically — you don't have to do anything special, just make sure draft_id is in your final message so the engine can extract it.

Prerequisite

This skill requires postall-agent to be installed in the engine container:

pip install postall-agent

If postall-agent is not found (run_bash returns "command not found"), tell the user: "This skill needs the postall-agent CLI installed. Run pip install postall-agent in the engine container and retry."