analysis-core
Analysis Core
Internal shared skill. Single source of truth for analysis-stage methodology used by PDCA fix workflows. Workflows keep their own orchestration (exits, manual/auto mode, OpenSpec/Jira artifact sinks, intentional divergences).
Prerequisite check: this skill declares strong dependencies in frontmatter. On load, verify each is available; if any is missing, abort and print the install command (
npx skills add FuDesign2008/open-skills -g --skill '*' --yes). No silent fallback.
Placeholder contracts
This skill is shared by workflows with different stage numbering. Never hardcode a workflow's stage numbers or titles here.
| Placeholder | Meaning | Who maps it |
|---|---|---|
{next-stage} |
Stage to enter after analysis exit gate / after instrumentation resolves | Referencing workflow, at the reference line (number + name) |
{root-cause step} / {impact-assessment step} / {upstream-eval step} |
Step numbers inside the analysis stage for known-issue-research |
Same reference line (existing known-issue-research contract) |
1. Temporary-change permission and rollback gate
Analysis is read-only by default. Analysis-assist edits only are allowed; they must be registered and fully rolled back before entering {next-stage}. Fix implementation belongs in the workflow's execution stage — not here.