hermes - 💡(How to fix) Fix Skill proposal: evidence-cited rationale archaeology for code history

Source pattern: cursor/plugins pstack at 21327be, especially the why skill.

Hermes adaptation:

• Add a Hermes-native "why/rationale archaeology" skill or mode for questions like "why was this built this way?".
• Start from a concrete code anchor: files, line ranges, symbols, recent commits, PR numbers, linked issues.
• Query evidence categories separately and in parallel where tools exist: git/gh, GitHub Issues/PRs, docs, notes, chat/email, observability, error tracking, analytics/MCP sources.
• Treat null results as evidence and report coverage gaps explicitly.
• Synthesize with confidence-calibrated language. Claims about intent need citations; inference must be labeled.

Why this is useful:

• Hermes often maintains older repos where implementation intent is scattered across PRs, issues, AGENTS.md, notes, and gateway/email history.
• Current coding workflows answer "what does it do" well, but can over-infer intent from code shape.
• This pattern is compatible with Hermes skills, GitHub tooling, MCP, and existing research/code-review skills. No Cursor runtime files are needed.

Acceptance criteria:

• New or extended built-in skill documents the code-anchor step, evidence-source matrix, null-result handling, and final answer format.
• The workflow works with only git + gh available, and opportunistically uses MCP/search/email/docs tools when configured.
• Output includes "sources searched", "what is directly evidenced", "what is inferred", and "open gaps".
• The skill cross-links systematic-debugging, github-code-review, native-mcp, and himalaya/google-workspace where appropriate.