Skill Name
root-cause-analysis
Programming Language
Other
Purpose
Provide a portable, evidence-backed root cause analysis workflow for generic Agent Skills hosts and Azure SRE Agent.
The workflow should investigate incidents, software defects, data failures, and process failures by forming falsifiable competing hypotheses, executing discriminating tests, preserving contradictory evidence, and applying an explicit causal completion gate before declaring a root cause.
Companion skill capabilities should support eligible statistical comparisons and reasoned assessment of how individual evidence items affect hypothesis plausibility without allowing either companion capability to set the final hypothesis disposition.
Prerequisites
An Agent Skills-compatible host.
The statistical-hypothesis-testing companion requires a statistical runtime that supports Barnard's exact test and Welch's independent-samples t-test.
Azure SRE mode requires the applicable Azure SRE connectors and read-only tools.
Generic hosts can operate on supplied evidence or approved evidence-expansion tools.
Requirements
- Support generic and Azure SRE host modes.
- Default to read-only investigation.
- Require explicit approval for production-altering, externally visible, or k evidence-altering actions.
- Maintain stable investigation, source, hypothesis, test, evidence, and likelihood records.
- Form falsifiable competing hypotheses and define predictions before testing.
- Distinguish Planned, Executed, Failed, and Blocked tests.
- Route eligible grouped observations to statistical-hypothesis-testing.
- Route selected evidence pairs to causal-evidence-likelihood without double-counting statistical evidence.
- Keep final hypothesis disposition and completion authority in the primary RCA workflow.
- Require all causal completion-gate criteria before returning Complete.
- Support Complete, Provisional, Inconclusive, Blocked, and Paused stop states.
- Return a resumable checkpoint for every non-Complete outcome.
- Include behavioral conformance coverage and synthetic regression cases for incident, data, software, and process failures.
- Publish generated reference documentation and include all three companion skills in plugin membership.
Skill Name
root-cause-analysis
Programming Language
Other
Purpose
Provide a portable, evidence-backed root cause analysis workflow for generic Agent Skills hosts and Azure SRE Agent.
The workflow should investigate incidents, software defects, data failures, and process failures by forming falsifiable competing hypotheses, executing discriminating tests, preserving contradictory evidence, and applying an explicit causal completion gate before declaring a root cause.
Companion skill capabilities should support eligible statistical comparisons and reasoned assessment of how individual evidence items affect hypothesis plausibility without allowing either companion capability to set the final hypothesis disposition.
Prerequisites
An Agent Skills-compatible host.
The statistical-hypothesis-testing companion requires a statistical runtime that supports Barnard's exact test and Welch's independent-samples t-test.
Azure SRE mode requires the applicable Azure SRE connectors and read-only tools.
Generic hosts can operate on supplied evidence or approved evidence-expansion tools.
Requirements