Best possible solution:

Make exact main consistent across schema validation, doctor pruning, route resolution, and configured binding compilation, then add focused route/doctor regressions plus redacted real behavior proof.

Do we have a high-confidence way to reproduce the issue?

Yes, source-level. A non-empty agents.list plus agentId: "main" hits the current schema/doctor predicates, and the PR head still leaves runtime routing through pickFirstExistingAgentId.

Is this the best way to solve the issue?

No. Preserving main in schema and doctor is necessary, but the best fix must also preserve the same target in route resolution and configured binding compilation.

Full review comments:

• [P1] Preserve explicit main in route resolution — src/config/zod-schema.ts:1286-1289
  Adding DEFAULT_AGENT_ID here lets documented agentId: "main" bindings load, but matched route bindings still call pickFirstExistingAgentId(). With agents.list containing only entries like codex and claude, main is absent from the lookup map and resolves to the configured default listed agent instead of the implicit main session, so add the runtime fix and route regression before merge.
  Confidence: 0.93

Overall correctness: patch is incorrect
Overall confidence: 0.91

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 233b48daaab4.

Label justifications:

• P1: The PR targets a documented startup/config-loss regression but can still produce wrong-agent channel delivery after relaxing validation.
• merge-risk: 🚨 compatibility: The diff changes config load and doctor repair behavior for existing bindings[] entries during upgrade.
• merge-risk: 🚨 message-delivery: Accepted main route bindings can still resolve to a listed agent instead of the implicit main agent.
• rating: 🧂 unranked krab: Overall readiness is 🧂 unranked krab; proof is 🧂 unranked krab and patch quality is 🧂 unranked krab.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR body lists tests/check/build output only; before merge, add redacted after-fix config-load, doctor, gateway, or route-resolution output and update the PR body for re-review.

PR surface:

Source +6, Tests +52. Total +58 across 4 files.

What I checked:

• PR diff scope: The branch changes schema validation, doctor compatibility pruning, and tests only; it does not change route resolution or configured binding compilation. (src/config/zod-schema.ts:1286, 5177472ea93c)
• Runtime route consumer still remaps binding targets: resolveAgentRoute passes matched binding targets through pickFirstExistingAgentId, so accepting agentId: "main" in schema does not preserve that target at runtime. (src/routing/resolve-route.ts:663, 233b48daaab4)
• Missing main falls back to configured default listed agent: When agents.list is non-empty and the requested normalized id is absent, pickFirstExistingAgentId returns the fallback default agent id. (src/routing/resolve-route.ts:168, 233b48daaab4)
• Default agent can be the first listed ACP agent: resolveDefaultAgentId chooses the first/default agents.list entry when the list is non-empty, which can turn an explicit main fallback binding into codex or another listed agent. (src/agents/agent-scope-config.ts:92, 233b48daaab4)
• Configured binding compiler shares the same helper: Configured binding compilation also calls pickFirstExistingAgentId, so the runtime target-preservation issue is not limited to direct route resolution. (src/channels/plugins/configured-binding-compiler.ts:83, 233b48daaab4)
• Documented affected config shape: The ACP docs show agents.list with codex and claude plus route bindings targeting agentId: "main" for Discord and Telegram fallback routing. Public docs: docs/tools/acp-agents.md. (docs/tools/acp-agents.md:423, 233b48daaab4)

Likely related people:

• yetval: Commit bbdff39b6a3f added normalized binding-agent validation and doctor pruning for missing agents.list ids in the central files touched by this PR. (role: introduced current validation and doctor behavior; confidence: high; commits: bbdff39b6a3f; files: src/config/zod-schema.ts, src/commands/doctor/shared/legacy-config-core-migrate.ts)
• Squirbie: PR history for a1cb8d50ba8d shows focused work on route binding behavior and resolve-route.ts, adjacent to the runtime path that still needs main preservation. (role: recent routing contributor; confidence: medium; commits: a1cb8d50ba8d; files: src/routing/resolve-route.ts)
• steipete: Recent history includes documentation/refactor touches around doctor compatibility helpers, routing helpers, and config schema normalization in the affected area. (role: recent adjacent contributor; confidence: medium; commits: 4fcc7537ffec, 9b30ff181c14, 00d8d7ead059; files: src/commands/doctor/shared/legacy-config-core-migrate.ts, src/routing/resolve-route.ts, src/config/zod-schema.ts)

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.