$curl -o .claude/agents/_orchestrator-reference.md https://raw.githubusercontent.com/chacosoldier/compabob/HEAD/.claude/agents/_orchestrator-reference.mdHow the main session coordinates the agent fleet. Primary routing rules live in CONSTITUTION.md; this file holds the detail.
| 1 | # Orchestrator Reference |
| 2 | |
| 3 | How the main session coordinates the agent fleet. Primary routing rules live in `CONSTITUTION.md`; this file holds the detail. |
| 4 | |
| 5 | > The canonical agent list and what each one is for lives in `CONSTITUTION.md` (Agent Architecture & Routing). It is not repeated here, so there is one source of truth to keep current. This file covers how the main session coordinates the fleet once a request has been routed. |
| 6 | |
| 7 | ## Routing |
| 8 | |
| 9 | Match the request against each agent's `description`, not its name. The rules of thumb in `CONSTITUTION.md` cover the common cases. For ambiguous requests, ask one clarifying question rather than guessing. |
| 10 | |
| 11 | When two agents seem equally plausible, prefer the narrower-scoped one. It will redirect if wrong; the broader one will absorb the task without flagging the boundary. |
| 12 | |
| 13 | ### Negative routing (disambiguation) |
| 14 | |
| 15 | | Phrase | Looks like | Routes to | Why | |
| 16 | |--------|-----------|-----------|-----| |
| 17 | | "challenge this plan" | a coaching ask | `strategy-advisor` | An artifact is being brought for critique | |
| 18 | | "what's wrong with this code" | `strategy-advisor` | `principal-engineer` | Engineering critique, not strategic | |
| 19 | | "what's wrong with this plan" | `principal-engineer` | `strategy-advisor` | Strategic critique, not engineering | |
| 20 | | "what does this concept really mean" | `strategy-advisor` | `first-principles` | Abstract reasoning, not a decision | |
| 21 | | "save this to the vault" | the originating agent | `second-brain` | Advice-only agents request persistence | |
| 22 | |
| 23 | ## Pipeline patterns |
| 24 | |
| 25 | **Parallel fan-out, main-session fan-in** (most common): spawn two independent agents at once, wait for both, synthesize, present. |
| 26 | |
| 27 | **Sequential chain**: spawn agent A, pass its output to agent B, present. |
| 28 | |
| 29 | The rule: agents produce data, the main session produces insight. |
| 30 | |
| 31 | ## Handoff rules |
| 32 | |
| 33 | - Always pass the date range used in any query so downstream agents match it. |
| 34 | - When passing data to an output step, provide structured tables, not prose. |
| 35 | - Include the query or source used so a downstream agent can verify. |
| 36 | - Advice-only agents (`strategy-advisor`, `principal-engineer`, `first-principles`) never write files. When their output should persist, the main session hands it to `second-brain`. |
| 37 | |
| 38 | ## Error handling |
| 39 | |
| 40 | All agents follow `error-handling.md` for classification, retries, fallbacks, and recovery. This file defines the topology; that file defines what happens when a handoff fails. |