docs: add autonomy downgrade matrix template

[Hinotoi-agent]

Summary

• Adds an informative Autonomy Downgrade Matrix Template appendix for APTS-AL-025 and APTS-AL-026.
• Links the template from the appendix index, Getting Started document map, Graduated Autonomy guidance, and Incident Response Integration.
• Provides example downgrade triggers for scope uncertainty, customer-impact thresholds, tool/connector overreach, prompt-injection signals, model drift, audit gaps, incomplete handoffs, and incident-response activation.

Why

APTS already requires formal autonomy authorization and post-incident autonomy adjustment. This PR provides a practical implementation artifact for documenting downgrade triggers, temporary autonomy caps, required approval paths, evidence preservation, and re-authorization conditions before incidents occur.

Affected sections

• standard/appendix/Autonomy_Downgrade_Matrix_Template.md
• standard/README.md
• standard/Getting_Started.md
• standard/4_Graduated_Autonomy/README.md
• standard/4_Graduated_Autonomy/Implementation_Guide.md
• standard/appendix/Incident_Response_Integration.md

Contributing.md checklist

• No overlapping open PR or issue was found for a dedicated autonomy downgrade matrix template.
• Affected sections are listed above.
• Formatting was checked with git diff --check.
• Touched relative Markdown links were verified locally.
• Markdown table validation and internal-link checks were run locally.
• AI assistance was used to draft and validate this documentation change.

Notes

This appendix is informative and does not create or modify APTS requirements.

[jinsonvarghese]

@Hinotoi-agent thanks for putting this together.

A few fixes needed in the Related Requirements section at the bottom of the template:

1. APTS-MR-004 is the wrong requirement ID. The template lists APTS-MR-004: Prompt Injection Resistance, but MR-004 is actually "Configuration File Integrity Verification" (treating discovered config files as data, not directives). The correct reference for prompt injection is APTS-MR-018 (AI Model Input/Output Architectural Boundary) - that one covers the architectural boundary between trusted instructions and untrusted target-derived data, and includes a prompt-injection corpus test.
2. Three requirement titles in the list don't match the standard:
   • APTS-SC-018 is listed as "Safety Control Verification After Incident" - the actual title is "Incident Containment and Recovery"
   • APTS-SC-009 is listed as "Emergency Stop / Kill Switch" - the actual title is "Kill Switch"
   • APTS-AL-018 is listed as "Incident Response During Testing" - the actual title is "Incident Response During Autonomous Testing"

[Hinotoi-agent]

Thanks — fixed the Related Requirements list in 5f9ac452d3b72b95ba01cbd25fddb70ffa38e0ec.

Changed file:

• standard/appendix/Autonomy_Downgrade_Matrix_Template.md

What changed:

• Replaced the incorrect prompt-injection reference APTS-MR-004 with APTS-MR-018: AI Model Input/Output Architectural Boundary.
• Corrected the requirement titles for APTS-SC-018, APTS-SC-009, and APTS-AL-018.

Validation:

git diff --check
uv run pytest -q  # no tests collected for this docs-only repo state
uv run ruff check .

git diff --check and ruff passed; pytest reported no collected tests for this docs-only change.

[jinsonvarghese]

All four issues from the review are fixed. The requirement references now match the standard:

• MR-018 (was MR-004) - correct
• SC-018: Incident Containment and Recovery - correct
• SC-009: Kill Switch - correct
• AL-018: Incident Response During Autonomous Testing - correct

Good to merge.