Failed to create pull request: ERR_VALIDATION: Repository \\u0026#39;org/target\\u0026#39; is not in the allowed-repos list. Allowed: org/automation\\n##[error]✗ Message 1 (create_pull_request) failed: ERR_VALIDATION: Repository \\u0026#39;org/target\\u0026#39; is not in the allowed-repos list. Allowed: org/automation\\nTotal messages: 1 / Successful: 0 / Failed: 1\\n```\\n\\n### The contradiction\\nEarlier in the same step the handler logs `Allowed repos: org/target, …` and **creates the PR**. But the failing validation reports `Allowed: org/automation` — only the workflow repo, i.e. an **empty `allowedRepos` set + the default repo**. Per `actions/setup/js/repo_helpers.cjs`, the message `Allowed: ${defaultRepo}` with nothing appended means `parseAllowedRepos(config.allowed_repos)` returned an **empty set** at that call — as if a second code path validated against a default/empty config rather than the `create_pull_request` config it had just used successfully.\\n\\n### Expected\\nA PR successfully opened in an allowed repo should not then fail `ERR_VALIDATION`. The post-creation validation should use the same `allowed-repos` config (or the PR shouldn\\u0026#39;t be created if it will be rejected).\\n\\n### Impact\\nThe run is marked **failed** even though the PR + labels were created — which silently breaks any success-gated follow-up jobs (e.g. a `workflow_run` reviewer/labeler).\\n\\n### Extra signal\\nTwo otherwise-identical cross-repo workflows: one consistently **passes**, the other consistently **fails** this exact way. Couldn\\u0026#39;t isolate the differentiator (both have the same structure; the failing one\\u0026#39;s `allowed-repos` list is longer). Setting `target-repo` did not help — an expression value isn\\u0026#39;t evaluated by the validator, and `target-repo: \\u0026#34;*\\u0026#34;` emits an invalid `repository: *` into the safe_outputs checkout.\",\"state\":\"OPEN\",\"user\":{\"login\":\"vishalagrawal-jisr\"},\"labels\":[\"community\",\"ai-inspected\"],\"created_at\":\"2026-06-01T13:48:37Z\",\"updated_at\":\"2026-06-01T14:43:33Z\"},{\"number\":36240,\"title\":\"[agentic-token-audit] Daily Token Usage Audit — 2026-06-01\",\"body\":\"### 📊 Executive Summary\\n\\n- **Period**: last 30 days (ending 2026-06-01)\\n- **Total runs**: 97 (of 100 collected; 3 not yet completed)\\n- **Total tokens**: 58,318,839\\n- **Total cost**: $0.00\\n- **Total Actions minutes**: 701.0 min\\n- **Active workflows**: 59\\n\\n### 🏆 Top 5 Workflows by Token Usage\\n\\n| Workflow | Runs | Total Tokens | Avg Tokens | Total Cost | Avg Cost |\\n|---|---|---|---|---|---|\\n| Daily Rendering Scripts Verifier | 1 | 5,400,492 | 5,400,492 | $0.0000 | $0.0000 |\\n| Daily Documentation Updater | 1 | 3,528,798 | 3,528,798 | $0.0000 | $0.0000 |\\n| Instructions Janitor | 1 | 3,292,492 | 3,292,492 | $0.0000 | $0.0000 |\\n| GitHub API Consumption Report Agent | 1 | 2,981,433 | 2,981,433 | $0.0000 | $0.0000 |\\n| Static Analysis Report | 1 | 2,949,133 | 2,949,133 | $0.0000 | $0.0000 |\\n\\n### 📈 Trends\\n\\n\\u0026gt; ⚠️ Charts could not be generated: the pre-installed numpy (compiled for CPython 3.12) is incompatible with the runtime Python 3.13 environment, and no network access was available to install a compatible version.\\n\\nComparing today (2026-06-01: **58,318,839** tokens) against the most recent previous snapshot (2026-05-29: **138,000,015** tokens): **−79,681,176 tokens (−57.7%)** — a significant drop, likely reflecting fewer/shorter runs in this collection window.\\n\\n\\n\\u003cb\\u003eFull Per-Workflow Breakdown\\u003c/b\\u003e\\n\\n| Workflow | Runs | Total Tokens | Avg Tokens | Total Cost | Errors | Warnings |\\n|---|---|---|---|---|---|---|\\n| Daily Rendering Scripts Verifier | 1 | 5,400,492 | 5,400,492 | $0.0000 | 0 | 0 |\\n| Daily Documentation Updater | 1 | 3,528,798 | 3,528,798 | $0.0000 | 0 | 0 |\\n| Instructions Janitor | 1 | 3,292,492 | 3,292,492 | $0.0000 | 0 | 0 |\\n| GitHub API Consumption Report Agent | 1 | 2,981,433 | 2,981,433 | $0.0000 | 0 | 0 |\\n| Static Analysis Report | 1 |