Best possible solution:

Land the narrow runtime-snapshot cron fix after normal maintainer review and required CI, keeping the regression test with the behavior change.

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

Yes, by source inspection: current main rebuilds cfgWithAgentDefaults from input.cfg, which can make Discord token resolution fall back to an unresolved SecretRef instead of the active runtime snapshot. I did not execute the regression test in this read-only review.

Is this the best way to solve the issue?

Yes, the PR uses the existing runtime snapshot selector before agent-default derivation rather than adding a new config path or Discord-specific workaround. The regression coverage targets the reported isolated Discord delivery failure.

AGENTS.md: found and applied where relevant.

Codex review notes: model gpt-5.5, reasoning high; reviewed against e844d1d6e56b.

Label changes:

• add P1: The PR fixes a reported regression where isolated cron Discord announce delivery can fail because resolved SecretRef credentials are not preserved.
• add rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🌊 off-meta tidepool and patch quality is 🐚 platinum hermit.
• add status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Not applicable: This maintainer-labeled PR includes focused local test proof; the external contributor real-behavior-proof gate is not applied to this protected maintainer workflow.

Label justifications:

• P1: The PR fixes a reported regression where isolated cron Discord announce delivery can fail because resolved SecretRef credentials are not preserved.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🌊 off-meta tidepool and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Not applicable: This maintainer-labeled PR includes focused local test proof; the external contributor real-behavior-proof gate is not applied to this protected maintainer workflow.

PR surface:

Source +16, Tests +50. Total +66 across 2 files.

What I checked:

• Root repository policy read and applied: Full root AGENTS.md was read; its maintainer-label protection and config/auth/message-delivery compatibility guidance shaped the review. No scoped AGENTS.md owns src/cron or src/config directly. (AGENTS.md:1, e844d1d6e56b)
• Current main still derives cron defaults from source config: Current main's prepareCronRunContext resolves default agent/config overrides from input.cfg and then spreads input.cfg into cfgWithAgentDefaults, which can make the config no longer match the runtime source snapshot after defaults are rebuilt. (src/cron/isolated-agent/run.ts:492, e844d1d6e56b)
• Runtime snapshot contract explains the failure mode: selectApplicableRuntimeConfig returns the runtime config only when the input config equals or matches the runtime source snapshot; a locally modified copy falls back to the input config. (src/config/runtime-snapshot.ts:209, e844d1d6e56b)
• Discord token resolution depends on the selected runtime config: Discord token resolution calls selectDiscordRuntimeConfig and reports configured_unavailable for unresolved account SecretRefs, matching the linked issue's observed error path. (extensions/discord/src/token.ts:59, e844d1d6e56b)
• Patch moves cron preparation onto the runtime snapshot: The PR adds resolveCronActiveRuntimeConfig and uses runtimeCfg for default agent resolution, agent overrides, agent defaults, and cfgWithAgentDefaults. (src/cron/isolated-agent/run.ts:492, 230cf7f49062)
• Regression test covers resolved Discord channel credentials: The PR sets a source config with an env SecretRef and a runtime config with a resolved Discord token, then asserts direct Discord delivery receives the runtime channels config. (src/cron/isolated-agent.direct-delivery-core-channels.test.ts:321, 230cf7f49062)

Likely related people:

• steipete: git blame ties the current cron preparation, runtime snapshot selector, and Discord token selection lines to Peter Steinberger, and the PR head is also authored by steipete. (role: current implementation and runtime-snapshot area contributor; confidence: high; commits: 4012ae4f426a, 230cf7f49062; files: src/cron/isolated-agent/run.ts, src/config/runtime-snapshot.ts, extensions/discord/src/token.ts)
• Vincent Koc: The current main tip is a cron-related fix by Vincent Koc, so he is an adjacent routing candidate if maintainers want another cron-area reviewer. (role: recent adjacent cron contributor; confidence: medium; commits: e844d1d6e56b; files: src/cron/service/timer.regression.test.ts, src/agents/session-suspension.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.

• Eggs appear after the PR passes real-behavior proof. It is here for vibes, not verdicts: it does not change labels, ratings, merge decisions, or automation.
• The shell reacts to review momentum: open follow-up work warms it up, re-review makes it wobble, and a clean final review lets it hatch.
• Hatchability usually comes from sufficient real-behavior proof, no blocking P0/P1/P2 findings, no security attention needed, and clean correctness. A merged PR is already final, so merge makes the egg hatchable independently.
• The hatch is seeded from this repository and PR number, so the same PR keeps the same creature; the reviewed head SHA can only change safe visual details.
• Rarity is just collectible sparkle: 🥚 common, 🌱 uncommon, 💎 rare, ✨ glimmer, and 🌈 legendary.