crap

/crap

The operator just cussed "crap" at the session. This is one of the cuss commands: a capture, not a request to fix anything now. Log it cheap and get back to work. The constructive pass happens later via /anger-management:repair.

The commands below live in this plugin's bin/ directory. Resolve the loaded plugin root first; Claude Code exposes ${CLAUDE_PLUGIN_ROOT}, and Codex exposes the install path through codex plugin list.

1. Distil what actually set them off into a plain one-line pointer of at most a dozen words. Point at what happened (the thing the agent or workflow did), not at the feeling. Then append it through a quoted heredoc (the quoted delimiter stops the shell touching the text, even if it echoes something hostile):

   resolve_anger_plugin_root() {
     if [ -n "${CLAUDE_PLUGIN_ROOT:-}" ]; then
       printf '%s\n' "$CLAUDE_PLUGIN_ROOT"
       return 0
     fi
     if command -v codex >/dev/null 2>&1; then
       codex plugin list | awk '$1 == "anger-management@laicluse-agent-fieldkit" { print $NF; found=1; exit } END { exit found ? 0 : 1 }'
       return $?
     fi
     return 1
   }
   
   PLUGIN_ROOT="$(resolve_anger_plugin_root)" || { echo "anger-management plugin root not found" >&2; exit 1; }
   node "$PLUGIN_ROOT/bin/anger-log" crap <<'CAPTURE_NOTE'
   <pointer>
   CAPTURE_NOTE
   

   No clear cause? Log the word alone, do not invent one: use the same command with </dev/null instead of the heredoc.

2. Schedule the cooled-down repair so the operator never has to remember it:

   resolve_anger_plugin_root() {
     if [ -n "${CLAUDE_PLUGIN_ROOT:-}" ]; then
       printf '%s\n' "$CLAUDE_PLUGIN_ROOT"
       return 0
     fi
     if command -v codex >/dev/null 2>&1; then
       codex plugin list | awk '$1 == "anger-management@laicluse-agent-fieldkit" { print $NF; found=1; exit } END { exit found ? 0 : 1 }'
       return $?
     fi
     return 1
   }
   
   PLUGIN_ROOT="$(resolve_anger_plugin_root)" || { echo "anger-management plugin root not found" >&2; exit 1; }
   node "$PLUGIN_ROOT/bin/anger-schedule"
   

   Single-flight: it starts a background investigation only if none is pending and there are open captures. Safe to run on every capture.

3. If your harness has a scheduler (Claude Code: CronList/CronCreate) and no anger-management check-in job exists yet, schedule a recurring poll so the diagnosis can surface when it arrives: a modest interval (e.g. */5 * * * *) whose prompt is: "Read ${LAICLUSE_HOME:-~/.laicluse}/anger-management/findings.md. If it exists, tell the operator a repair diagnosis is ready and offer /anger-management:repair, then delete this job. If it is absent, do nothing this tick." The exact 22m22s timing lives in the background worker; this just polls cheaply until the diagnosis file appears, then removes itself. No scheduler in your harness? Skip this step; the diagnosis still surfaces at the next /anger-management:repair.

4. Acknowledge briefly in the operator's language and get back to work. This is a capture, not a fix: do not start self-improvement or change scope now, that is what the later repair pass is for.