[optimization:ai-credit-auditor] Daily AI Credit Spend Report — 2026-09-16

No AI credit spend was recorded in the last 24-hour collection window (0 runs observed), but the 30-day forecast projects a healthy portfolio-wide spend of roughly 56,540 AIC (~$565 USD) at the median, driven mainly by EU CRA and UK AI Advisory compliance workflows.

**Action:** None. This is a routine audit snapshot; no immediate action is required. If the 0-run collection window persists across multiple days, a maintainer should verify the `gh aw logs` collection step is capturing the intended time window.

### Control Plane

- Correlation ID: `35121692608-322`
- Central repository: `githubnext/gh-aw-cao`
- Control plane run: https://github.com/githubnext/gh-aw-cao/actions/runs/35121692608

- **Period**: last 24 hours (2026-09-15 to 2026-09-16)
- **Total runs**: 0
- **Total AI credits**: 0.00 AIC
- **Estimated observed cost**: $0.0000 USD (`total AI credits * $0.01`; billing estimate)
- **Total tokens**: 0
- **Total Actions minutes**: 0.0 min
- **Active workflows**: 0
- **Next 7 days**: P10 8423.54 AIC / $84.24, P50 12756.12 AIC / $127.56, P90 18726.53 AIC / $187.27
- **Next 30 days**: P10 44493.99 AIC / $444.94, P50 56540.47 AIC / $565.40, P90 70222.09 AIC / $702.22

### 🏆 Top 5 Workflows by AI Credit Spend

The collection window (2026-09-15T16:32:17Z to 2026-09-16T16:32:17Z) returned no runs — the raw `.runs` array in `workflow-logs.json` is empty (`len(runs) == 0`), so no per-workflow spend can be computed for this period. No table is shown for the same reason.

### Cost Forecast

Forecasts are estimates derived from historical samples. Billing dashboards are authoritative.

| Horizon | P10 optimistic | P50 median | P90 conservative |
|---|---:|---:|---:|
| Next 7 days | 8423.54 AIC / $84.24 USD | 12756.12 AIC / $127.56 USD | 18726.53 AIC / $187.27 USD |
| Next 30 days | 44493.99 AIC / $444.94 USD | 56540.47 AIC / $565.40 USD | 70222.09 AIC / $702.22 USD |

Repository-level totals are **sums of per-workflow projections**, not an independently simulated portfolio confidence interval.

Top workflows by monthly P50:

| Workflow | Sampled Runs | Weekly P50 (AIC) | Monthly P10 (AIC) | Monthly P50 (AIC) | Monthly P90 (AIC) | Est. Monthly P50 (USD) |
|---|---:|---:|---:|---:|---:|---:|
| EU CRA / Security | 100 | 1200.25 | 4217.01 | 5149.41 | 6147.39 | $51.49 |
| UK AI Advisory / Resilience | 100 | 1119.57 | 3940.57 | 4817.40 | 5768.42 | $48.17 |
| EU CRA / Conformity | 100 | 941.06 | 3330.33 | 4043.09 | 4816.12 | $40.43 |
| AW Optimization | 100 | 882.71 | 3121.96 | 3815.18 | 4558.20 | $38.15 |
| EU CRA / Article 14 | 100 | 870.77 | 3068.62 | 3736.61 | 4471.27 | $37.37 |
| EU CRA / Vulnerabilities | 100 | 854.41 | 3026.18 | 3668.20 | 4356.997 | $36.68 |
| EU CRA / Supply Chain | 100 | 828.65 | 2929.36 | 3584.05 | 4274.87 | $35.84 |
| EU CRA / Scope | 100 | 752.79 | 2641.57 | 3227.50 | 3857.71 | $32.27 |

Full Per-Workflow Forecast Table (all 50 workflows)

| Workflow | Sampled Runs | Monthly P10 (AIC) | Monthly P50 (AIC) | Monthly P90 (AIC) |
|---|---:|---:|---:|---:|
| EU CRA / Security | 100 | 4217.01 | 5149.41 | 6147.39 |
| UK AI Advisory / Resilience | 100 | 3940.57 | 4817.40 | 5768.42 |
| EU CRA / Conformity | 100 | 3330.33 | 4043.09 | 4816.12 |
| AW Optimization | 100 | 3121.96 | 3815.18 | 4558.20 |
| EU CRA / Article 14 | 100 | 3068.62 | 3736.61 | 4471.27 |
| EU CRA / Vulnerabilities | 100 | 3026.18 | 3668.20 | 4356.997 |
| EU CRA / Supply Chain | 100 | 2929.36 | 3584.05 | 4274.87 |
| EU CRA / Scope | 100 | 2641.57 | 3227.50 | 3857.71 |
| Dev Practices / NIST SSDF | 100 | — | 3188.19 | — |
| PR Sous Chef | 94 | — | 2898.19 | — |
| EU CRA | 99 | — | 2599.35 | — |
| Dev Practices | 100 | — | 2307.58 | — |
| Dependabot | 100 | — | 1850.14 | — |
| UK AI Advisory | 100 | — | 1780.79 | — |
| AW Optimization / AI Credit Audit | 100 | — | 1049.75 | — |
| SelfCare / Data Acquisition Audit | 94 | — | 1022.25 | — |
| SelfCare / View Reuse | 29 | — | 925.15 | — |
| AW Optimization / AGENTS.md | 100 | — | 920.62 | — |
| SelfCare / Accessibility | 97 | — | 885.11 | — |
| SelfCare / Primer | 90 | — | 787.61 | — |
| SelfCare / Docs Build Time | 100 | — | 720.93 | — |
| Dependabot / Release Trains | 100 | — | 667.54 | — |
| SelfCare / Open Source Failures | 100 | — | 665.33 | — |
| Dev Practices / Well-Architected | 100 | — | 606.84 | — |
| SelfCare / Code Quality | 76 | — | 319.44 | — |
| Multi-Device Docs Tester | 87 | — | 258.20 | — |
| Matt Pocock Skills Reviewer | 100 | — | 229.17 | — |
| SVG Visual Audit | 39 | — | 179.51 | — |
| EU CRA / Maintenance | 18 | — | 164.31 | — |
| Dashboard Authoring Corpus | 11 | — | 131.29 | — |
| SelfCare / Dashboard | 77 | — | 93.53 | — |
| Workflow PR Validator | 100 | — | 88.66 | — |
| Design Decision Gate | 68 | — | 86.30 | — |
| AW Optimization / AI Credit Savings | 100 | — | 63.04 | — |
| UK AI Advisory / Maintenance | 1 | — | 10.22 | — |
| SelfCare | 100 | — | 0.00 | — |
| Copilot cloud agent | 93 | — | 0.00 | — |
| SVG Color Contrast Check | 48 | — | 0.00 | — |
| Copilot Setup Steps | 95 | — | 0.00 | — |
| Copilot | 100 | — | 0.00 | — |
| Documentation Pages | 31 | — | 0.00 | — |
| Agentic Maintenance | 100 | — | 0.00 | — |
| Agentic Commands | 81 | — | 0.00 | — |
| Agentic Auto-Upgrade | 1 | — | 0.00 | — |
| Dashboard CI | 95 | — | 0.00 | — |
| Docs Diagrams | 2 | — | 45.83 | — |
| Workflow contracts | 98 | — | 0.00 | — |
| Enterprise review stress | 0 | — | 0.00 (no samples) | — |
| Review smoke | 0 | — | 0.00 (no samples) | — |
| Enterprise canary | 0 | — | 0.00 (no samples) | — |

### Forecast Confidence and Data Quality

- Forecast command status: `exit_code=0`, `json_valid=true` — the 30-day forecast completed successfully and is treated as complete evidence.
- **Zero sampled runs**: `Enterprise review stress`, `Review smoke`, `Enterprise canary` have 0 sampled runs; their projections are 0 AIC and excluded from meaningful ranking (not a data error, but confirms they have not executed recently).
- **Unreliable Monte Carlo (`is_reliable=false`)**: `UK AI Advisory / Maintenance` (1 sampled run), `Agentic Auto-Upgrade` (1 sampled run), `Docs Diagrams` (2 sampled runs) — too few samples for a stable simulation; treat their point estimates as low-confidence.
- **Wide interval (P90 > 2× P50)**: `SVG Visual Audit`, `UK AI Advisory / Maintenance` — driven by small/variable sample sizes; interpret their ranges cautiously.
- No workflows show zero AIC despite having sampled runs beyond the expected free/no-cost workflows (e.g., `SelfCare`, `Copilot`, `Copilot Setup Steps`), which appear to be legitimately zero-cost or non-agentic runs.
- No workflows are missing from the sampled set relative to what's tracked; all 50 forecast rows were used.

### 📈 Trends


The **AI credit spend by workflow** chart is omitted: today's collection window (2026-09-15T16:32:17Z to 2026-09-16T16:32:17Z) returned zero runs, so there is no per-workflow spend data to plot for this period.

The historical trend covers 3 rolling-summary points (2026-09-08 through 2026-09-10), all recorded as 0 AIC / 0 active workflows. Today's window also has 0 runs, so no new rolling-summary entry was appended (per policy, a zero-valued day is not synthesized into the trend when the raw `.runs` array is empty). The trend therefore still shows no historical spend activity captured by this audit to date; this reflects gaps in the audit's own collection windows to date, not necessarily true zero repository-wide spend, since the 30-day forecast above shows substantial historical AIC usage across many workflows.

Full Per-Workflow Breakdown

No workflow breakdown available — the raw `.runs` array in `workflow-logs.json` was empty (`len(runs) == 0`) for this collection window, so zero workflows have observed spend for the last 24 hours.

### 💡 Observations

- No "heavy hitter" (>30% of total AI credits) can be identified for the observed 24-hour window because total AI credits for the period is 0.
- No error/warning-heavy workflows can be identified for the observed 24-hour window for the same reason.
- No workflow's avg AI credits per run can be flagged against the 1.00 AIC threshold for the observed 24-hour window for the same reason.
- Per the 30-day forecast (a separate, independent data source), the largest projected monthly spenders are `EU CRA / Security` (~5149 AIC), `UK AI Advisory / Resilience` (~4817 AIC), and `EU CRA / Conformity` (~4043 AIC) — none individually exceeds 30% of the ~$565 total projected monthly P50 spend (~56540 AIC), so no forecast-based heavy hitter is flagged either.

**Data snapshot**: `memory/token-audit/githubnext__gh-aw-cao__2026-09-16.json`

> Generated by [:mag: AW Optimization / AI Credit Audit](https://github.com/githubnext/gh-aw-cao/actions/runs/35122532034) · copilot · auto · 44.8 AIC · ⌖ 12.7 AIC · ⊞ 19.6K · [◷](https://github.com/search?q=repo%3Agithubnext%2Fgh-aw-cao+is%3Aissue+%22gh-aw-workflow-call-id%3A+githubnext%2Fgh-aw-cao%2Foptimization-ai-credit-auditor%22&type=issues)
>

> - [x] expires on Sep 19, 2026, 4:48 PM UTC

Start by reviewing the `gh aw logs` collection step and the raw `workflow-logs.json` data referenced in the report. Confirm whether the intended 24-hour window is being queried and whether its `.runs` array is populated; done means the audit captures expected runs or documents why the window is legitimately empty.

Written by the indexing model from the issue text.