Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Implement secret-safe HTTP headers for MCP transport in gh-aw.
.claude/skills/github-http-mcp-headers/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 45% | 0% |
Use this reference for HTTP MCP header secret support in the copilot engine.
When HTTP MCP headers include GitHub Actions secrets, mcp-config.json must:
${{ secrets.DD_API_KEY }})markdownon: workflow_dispatch: permissions: contents: read engine: copilot mcp-servers: datadog: type: http url: "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp" headers: DD_API_KEY: "${{ secrets.DD_API_KEY }}" DD_APPLICATION_KEY: "${{ secrets.DD_APPLICATION_KEY }}" DD_SITE: "${{ secrets.DD_SITE || 'datadoghq.com' }}" allowed: - search_datadog_dashboards - search_datadog_slos - search_datadog_metrics - get_datadog_metric # Datadog Dashboard Search Search for Datadog dashboards and provide a summary.
json{ "mcpServers": { "datadog": { "type": "http", "url": "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp", "headers": { "DD_API_KEY": "${DD_API_KEY}", "DD_APPLICATION_KEY": "${DD_APPLICATION_KEY}", "DD_SITE": "${DD_SITE}" }, "tools": [ "search_datadog_dashboards", "search_datadog_slos", "search_datadog_metrics", "get_datadog_metric" ], "env": { "DD_API_KEY": "\\${DD_API_KEY}", "DD_APPLICATION_KEY": "\\${DD_APPLICATION_KEY}", "DD_SITE": "\\${DD_SITE}" } } } }
yamlenv: DD_API_KEY: ${{ secrets.DD_API_KEY }} DD_APPLICATION_KEY: ${{ secrets.DD_APPLICATION_KEY }} DD_SITE: ${{ secrets.DD_SITE || 'datadoghq.com' }} COPILOT_GITHUB_TOKEN: ${{ secrets.COPILOT_GITHUB_TOKEN }} # ... other env vars
GH_AW_MCP_CONFIG is intentionally NOT in the YAML env: block — it is exported from the run script (export GH_AW_MCP_CONFIG="$HOME/.copilot/mcp-config.json") so $HOME is resolved at runtime. GitHub Actions does not shell-expand env: values, so the path must be set via export to work on self-hosted/containerized runners where HOME is not /home/runner.
${{ secrets.VAR_NAME }} patterns${{ secrets.VAR || 'default' }}${{ secrets.DD_API_KEY }} to ${DD_API_KEY}\${VAR_NAME})${{ secrets.* }} syntaxAll tests pass ✓
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→pass | 11,098 | 6,420 | -42% | 1 | 1 | 0% | 1,982 | 2,605 | +31% | 0 | 0 | — |
case-01 | fail→fail | 9,304 | 7,468 | -20% | 1 | 1 | 0% | 1,581 | 2,788 | +76% | 0 | 0 | — |
case-02 | fail→pass | 7,002 | 6,482 | -7% | 1 | 1 | 0% | 1,292 | 2,807 | +117% | 0 | 0 | — |
case-03 | fail→pass | 13,382 | 5,470 | -59% | 1 | 1 | 0% | 2,411 | 2,422 | +0% | 0 | 0 | — |
case-04 | pass→pass | 6,267 | 7,843 | +25% | 1 | 1 | 0% | 1,162 | 2,786 | +140% | 0 | 0 | — |
case-05 | pass→pass | 10,350 | 6,827 | -34% | 1 | 1 | 0% | 1,754 | 2,502 | +43% | 0 | 0 | — |
case-06 | pass→pass | 8,587 | 6,360 | -26% | 1 | 1 | 0% | 1,486 | 2,400 | +62% | 0 | 0 | — |
case-07 | fail→pass | 13,308 | 5,666 | -57% | 1 | 1 | 0% | 2,408 | 2,395 | -1% | 0 | 0 | — |
case-08 | pass→pass | 11,998 | 3,733 | -69% | 1 | 1 | 0% | 1,960 | 1,867 | -5% | 0 | 0 | — |
case-09 | fail→pass | 11,091 | 3,296 | -70% | 1 | 1 | 0% | 1,956 | 1,773 | -9% | 0 | 0 | — |
case-10 | fail→pass | 9,792 | 6,223 | -36% | 1 | 1 | 0% | 1,756 | 2,544 | +45% | 0 | 0 | — |
case-11 | fail→pass | 14,818 | 6,906 | -53% | 1 | 1 | 0% | 2,379 | 2,610 | +10% | 0 | 0 | — |
case-13 | fail→pass | 9,671 | 4,096 | -58% | 1 | 1 | 0% | 1,644 | 1,993 | +21% | 0 | 0 | — |
case-14 | fail→pass | 8,077 | 2,867 | -65% | 1 | 1 | 0% | 1,253 | 1,663 | +33% | 0 | 0 | — |
case-15 | pass→pass | 9,009 | 5,402 | -40% | 1 | 1 | 0% | 1,578 | 2,309 | +46% | 0 | 0 | — |
case-16 | pass→pass | 6,851 | 4,261 | -38% | 1 | 1 | 0% | 1,205 | 1,988 | +65% | 0 | 0 | — |
case-17 | fail→pass | 13,235 | 3,781 | -71% | 1 | 1 | 0% | 2,252 | 1,870 | -17% | 0 | 0 | — |
case-18 | fail→pass | 11,020 | 2,985 | -73% | 1 | 1 | 0% | 1,703 | 1,826 | +7% | 0 | 0 | — |
case-19 | fail→pass | 11,548 | 3,173 | -73% | 1 | 1 | 0% | 2,246 | 1,853 | -17% | 0 | 0 | — |
case-20 | fail→pass | 11,455 | 4,733 | -59% | 1 | 1 | 0% | 1,932 | 2,091 | +8% | 0 | 0 | — |
case-21 | fail→pass | 13,854 | 8,836 | -36% | 1 | 1 | 0% | 2,255 | 2,940 | +30% | 0 | 0 | — |
case-22 | pass→pass | 6,914 | 3,123 | -55% | 1 | 1 | 0% | 1,158 | 1,883 | +63% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +59 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.