[maestro] Ensure objective_id is set for codex worker operating-chat requests

[haasonsaas]

Summary

Fixes Sentry PLATFORM-5 — operating_chat_objective_id_required: worker error details redacted (culprit codex_worker.operating_chat.operating_chat).

Source-of-truth PR: https://github.com/evalops/maestro-internal/pull/2845

Root cause

The Platform "codex worker" operating-chat / tool-execution endpoint requires objective_id on the request linkage for the platform-agent-runtime surface, but this client treated objective_id as optional end-to-end. Compounding it, MAESTRO_OBJECTIVE_ID was not in the worker's trusted-tool env allowlist, so the objective context could be stripped before it ever reached the request linkage — producing the redacted server-side rejection.

Fix

• src/utils/shell-env.ts — add MAESTRO_OBJECTIVE_ID and MAESTRO_EVALOPS_OBJECTIVE_ID to PLATFORM_TRUSTED_TOOL_ENV_ALLOWLIST so the objective survives into worker/trusted-tool subprocesses.
• src/platform/tool-execution-client.ts — resolve objectiveId from the linkage, falling back to MAESTRO_OBJECTIVE_ID / MAESTRO_EVALOPS_OBJECTIVE_ID, so a configured objective always reaches the wire. Added a pre-flight guard scoped to the platform-agent-runtime surface that fails fast with a clear, unredacted error when objective_id is still missing.
• src/telemetry/maestro-event-bus.ts — read objective_id from both env var names for consistency with auth.ts.
• test/platform/tool-execution-client.test.ts — added coverage for linkage passthrough, env fallback (both vars), and the worker-surface guard.

Validation

• bun run bun:lint — pass (biome + codex/platform conformance validators).
• bunx vitest --run test/platform/tool-execution-client.test.ts test/platform/agent-runtime-client.test.ts — pass (28 tests).

🔗 Slack thread: https://evalops.slack.com/archives/C0B8Y1J5U1H/p1782505086813659?thread_ts=1782505074.141379&cid=C0B8Y1J5U1H

---

Generated by Claude Code

[github-actions]

This PR changes mirrored Maestro source files in the public repo, but it does not link the matching private source-of-truth PR.

Add one of these to the PR body, then re-run the check:

• https://github.com/evalops/maestro-internal/pull/<number>
• evalops/maestro-internal#<number>
• maestro-internal#<number>

Mirrored files touched:

• packages/tui-rs/src/tools/shell_env.rs
• src/platform/tool-execution-client.ts
• src/telemetry/maestro-event-bus.ts
• src/utils/shell-env.ts
• test/platform/tool-execution-client.test.ts

[cursor]

PR Summary

Medium Risk
Changes tool-execution linkage and worker env allowlists on a critical Platform integration path; misconfiguration now fails fast instead of opaque server errors.

Overview
Fixes PLATFORM-5 by ensuring objective_id reaches Platform tool-execution requests for the codex worker (platform-agent-runtime) surface instead of being dropped or omitted.

Trusted worker env: MAESTRO_OBJECTIVE_ID and MAESTRO_EVALOPS_OBJECTIVE_ID are added to the platform trusted-tool allowlist in Node (shell-env.ts) and Rust (tui-rs shell_env.rs) so filtered shell subprocesses still see objective context.

Wire-up: Tool linkage and telemetry now resolve objective from linkage first, then MAESTRO_OBJECTIVE_ID / MAESTRO_EVALOPS_OBJECTIVE_ID (bridge, event bus). executeToolWithPlatform applies the same resolution when normalizing requests and fails fast with a clear error if the worker surface has no objective—before any HTTP call.

Tests: Coverage for env fallbacks, linkage passthrough, worker-surface guard, and bridge linkage including MAESTRO_EVALOPS_OBJECTIVE_ID.

^{Reviewed by Cursor Bugbot for commit 6bce0a8. Bugbot is set up for automated code reviews on this repo. Configure here.}

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[devin-ai-integration]

Devin Review found 0 new potential issues.

[devin-ai-integration]

Devin Review found 2 new potential issues.