Add: SOUL.md intro & 3 approaches — sources ingested

references/agent-preferences.md

@@ -0,0 +1,20 @@
+# Wiki Agent — Style & Workflow Preferences (2026-09-25 session)
+
+Captured from session with user Alister (UK ~58, direct/concise, never invent creds, saves docs to /opt/data/). Applied via wiki-ingest pipeline (user-owned skill — recommend `hermes curator adopt wiki-ingest`).
+
+## Preferences embedded
+- **Brevity / directness:** Replies match length to ask weight; no filler ("Great question", "I'd be happy to"); start with answer; no restating the question. Confirmed by user's own `SOUL.md` content.
+- **Verifiable claims:** When grounding in source, cite; when inferring, label. Use exact paths, exact identifiers.
+- **Failure honesty:** Say "I don't know" when unsure; distinguish recall vs reasoning.
+
+## Techniques captured (usable next session)
+- SOUL.md file case-sensitive search (`find` / `-iname`) + profile isolation check (`profiles/wiki/` vs global).
+- Raw-source → wiki pipeline verified with schema (frontmatter + `raw/articles/` + `index.md` + `log.md` + Gitea push with embedded PAT).
+- Three-approach comparison method: collect official docs + community tutorial + feature-overview source; write comparative concept page.
+- No inventing of credentials (PAT read live from `/opt/data/.secrets/`; never hardcoded).
+
+## Skill status
+- `wiki-ingest` (user-owned): skill saved with `references/soul-md-research-summary.md`; SKILL.md patch refused (protected — user-owned). Recommendation: `hermes curator adopt wiki-ingest` to unlock autonomous patches.
+- `shopping`: consulted, not modified.
+- `hermes-agent`: loaded, not modified.
+- Overlap note: `wiki-ingest` and `llm-wiki` (if exists) may overlap on wiki-ingest steps — background curator to consolidate if both present.

references/soul-md-approach-community.md

@@ -0,0 +1,16 @@
+# Raw Source: SOUL.md — Approach B (Community / Template-Based)
+
+Source: https://www.stanza.dev/courses/hermes-personality/soul-md/hermes-personality-soul-identity-layer
+Source: https://blog.devgenius.io/how-to-play-with-soul-md-in-hermes-agent-135d1a36c9f9
+Type: community tutorials / educational guides
+Confidence: medium-high (secondary sources referencing official docs)
+Date collected: 2026-09-25
+
+## Summary
+Community tutorials emphasize that SOUL.md is an "identity layer" — separate from preferences (USER.md), operating procedures (AGENTS.md), tool notes (TOOLS.md), and memory (MEMORY.md). The recommended structure: Identity → Style → Technical Preferences → Ethical Boundaries. Keep it under ~1KB / 2,000 tokens; it competes for context window space.
+
+## Key differences from official guidance
+- More structured section headings encouraged (Identity, Communication Style, Technical Preferences, Ethical Boundaries).
+- Explicit size guideline: under 1KB preferred; 2,000 token cap enforced by framework.
+- More emphasis on per-project workspace-level SOUL.md overrides (workspace SOUL.md wins over global).
+- Stronger focus on "hard limits as prompt-injection defense" — using negative rules ("Never do X") as security layer.

references/soul-md-approach-official.md

@@ -0,0 +1,29 @@
+# Raw Source: SOUL.md Introduction & Methods — Approach A (Official / Pragmatic)
+
+Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/personality
+Source: https://hermes-agent.nousresearch.com/docs/guides/use-soul-with-hermes
+Type: official documentation (Nous Research / Hermes Agent)
+Confidence: high (primary source)
+Date collected: 2026-09-25
+
+## Summary
+SOUL.md is the agent's primary identity file at `~/.hermes/SOUL.md` (or `$HERMES_HOME/SOUL.md`). It lives at slot #1 of the system prompt — the first text the model sees before any user message. It defines tone, personality, communication style, and how the agent handles uncertainty, disagreement, and ambiguity.
+
+## Key rules
+- Load path: ONLY from `HERMES_HOME` (never from cwd).
+- Never overwritten by Hermes; starter file auto-seeded if missing.
+- Empty/missing file → falls back to built-in default identity.
+- Scanned for prompt injection before inclusion; content injected verbatim after truncation.
+- Stable identity: durable voice, not project-specific rules.
+
+## What goes in SOUL.md (official recommendation)
+- Identity / who the agent is
+- Style / how it communicates
+- Avoid / stylistic prohibitions
+- Defaults / behavior under ambiguity
+
+## What does NOT go in SOUL.md
+- Project conventions → AGENTS.md
+- File paths, repo rules → AGENTS.md
+- Tool preferences / commands → AGENTS.md or config.yaml
+- Memory / user context → MEMORY.md / USER.md

references/soul-md-approach-personality.md

@@ -0,0 +1,16 @@
+# Raw Source: SOUL.md — Approach C (Persona / Personality-Driven)
+
+Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/personality (built-in personalities / /personality overlay section)
+Source: https://hermes-agent.nousresearch.com/docs/guides/use-soul-with-hermes (SOUL.md vs /personality comparison)
+Type: official feature docs (personality system)
+Confidence: high
+Date collected: 2026-09-25
+
+## Summary
+SOUL.md defines the durable baseline identity. `/personality` provides session-level overlays. The framework supports custom personality presets defined in `config.yaml` (`agent.personalities`). Each profile can have its own `SOUL.md` (via profile directory under `profiles/<name>/`).
+
+## Key differences
+- SOUL.md = durable identity; /personality = temporary mode switch.
+- Profile-level SOUL.md isolation: `hermes profile create --clone` creates independent persona files.
+- Personality presets defined in YAML (`agent.personalities:`) apply to the whole session without touching SOUL.md file.
+- Personality names are stored in display metadata; personal files stay in SOUL.md.