dsh-advisor

Usage

Once installed and enabled, the advisor observes every session. Control it per session with the /advisor command (available when a command registry is composed):

/advisor            toggle the advisor for this session
/advisor on         enable the advisor for this session
/advisor off        disable the advisor for this session
/advisor status     show state, model, runtime status, pending count, last activity

/advisor on|off|toggle are session-scoped and ephemeral: they flip a per-session override, never the persisted config. Enabling a session whose config lacks provider/model starts no model call — /advisor status (and the /advisor on reply) shows the gate reason.

/advisor on is also the manual recovery path: a session advisor paused by a quota/rate-limit (quota_exhausted — KD-5 has no auto-resume timer) resumes in place, and a halted advisor (permanent model error, e.g. invalid credentials) is rebuilt fresh for the session.

The advisor reviews on a dual-mode trigger, depending on the session shape:

• Standard stepped sessions — after each stepped primary turn that ends normally (completed, max-tokens, or error), the advisor reviews the incremental transcript delta.
• Agentic / harness sessions (never emit turn/end) — after each completed agent reply round: when a new human input arrives (inbox-spliced input included) after an unreviewed assistant increment, the advisor reviews that increment.

Either way the advisor emits at most one note per review, ranked by severity:

• nit — a minor style, clarity, or quality suggestion; delivered via agent.inject (non-waking, consumed at the next pre-step boundary).
• concern — a material risk or clearly better direction to weigh before continuing; delivered via agent.steer (waking), subject to the immuneTurns cooldown.
• blocker — continuing clearly wastes work (contradicts an explicit user instruction, going in circles, fundamentally unsound); delivered via agent.steer.

Injected advice appears in the session stream as a user-role message carrying the advisor source kind and self-describing content, e.g.:

[advisor:concern] extract the helper into a module and unit-test it

The [advisor:{severity}] prefix is the only cue the primary model gets about how to treat it — the primary system prompt never mentions advisories. Advisor messages are excluded from later advisor deltas, so the advisor never reads its own advice back.