| name | agent-status-reporter |
| description | Report current agent/session status and system health — skill-loop run state, scheduler/memory counters, and active context. Use when the user runs /status in the abi REPL or asks what the agent is doing or for system health. |
agent-status-reporter
Backs the /status slash command in the agent tui REPL
(.opencode.json slash_commands.status). Reports a compact, read-only
snapshot:
What to report
- Skill-loop run state — read the last entry of
.skill-telemetry/runs.jsonl and .skill-telemetry/.sessions/ for the
active session id + status. Summarize the latest run's status + counts; do
not dump the file.
- Scheduler / memory — run
abi scheduler status (one-shot
self-terminating probe): report counters + attached MemoryTracker stats +
the always-on telemetry block. The probe is a no-op, so memory counters read
0 by design — say so; do not fabricate load.
- Context — give one line: loaded files + whether SEA mode is on (from
sea-learning-controller). Defer the full context view to
context-state-reporter (/context).
Rules
- Read-only. Never edit source, telemetry, or session files.
- No performance / accuracy claims without a repo benchmark.
- If a telemetry file is missing, report "not initialized" — do not invent.
