| name | freeqcc-launch |
| description | Launch a freeqcc agent — a freeq-DM-controllable Claude Code session with a cryptographic did:key identity, owner-DID-gated access, and a declarative provenance cert. Use when the user wants to start their agent and become reachable via Bluesky DM.
|
freeqcc launch
The freeqcc CLI runs the agent. Your job here:
-
Run freeqcc launch --detach (background mode).
- First-time users will be prompted in the terminal where you
run the command for their AT Protocol handle and a bot nick.
Make sure they know to look there if the command appears to
hang — the prompts library doesn't echo to Claude Code's
output stream.
- On subsequent launches the daemon starts immediately because
handle + nick are already persisted in
~/.freeqcc/.
-
After launch returns, run freeqcc status to read back the live
state and report:
- the bot's IRC nick + did:key DID
- the owner DID we'll only respond to
- provenance verified status (expected:
verified=false,
reason "Cert has no signature; declarative only" — that's
intentional in v1.0)
-
Tell the user how to DM it: from any freeq-connected client
signed in as the owner handle, send a Bluesky DM to the bot's
nick. The agent will dispatch to a persistent claude -p
session and reply.
If freeqcc isn't on PATH, ask the user to install:
cd /path/to/freeq/freeqcc && npm install && npm link
If the user wants to stop the agent later, the matching skills are
/freeqcc-status and /freeqcc-stop.