fix: split SKILL.md for Hermes Agent 100KB compatibility

.clawhubignore

@@ -49,6 +49,7 @@ CONCEPTS.md
 CONFIGURATION.md
 CONTRIBUTORS.md
 HERMES_SETUP.md
+HERMES_CRON_SETUP.md
 release-notes.md
 SKILL-original.md
 SPEC.md

.skillignore

@@ -48,6 +48,7 @@ CONCEPTS.md
 CONFIGURATION.md
 CONTRIBUTORS.md
 HERMES_SETUP.md
+HERMES_CRON_SETUP.md
 release-notes.md
 SKILL-original.md
 SPEC.md

HERMES_CRON_SETUP.md

@@ -0,0 +1,38 @@
+# Hermes Cron Setup for last30days
+
+> The SessionStart hook (`hooks/hooks.json`) is Claude Code-specific and does not fire on Hermes.
+> Use a Hermes cron job as the equivalent auto-start mechanism.
+
+## What the SessionStart hook does
+
+The Claude Code hook runs `hooks/scripts/check-config.sh` on every session start. It:
+1. Creates `$LAST30DAYS_MEMORY_DIR` (defaults to `~/Documents/Last30Days`)
+2. Shows configuration status (active sources, last run summary)
+3. Detects first-run and shows welcome message
+
+## Hermes equivalent: cron job
+### Directory creation
+The engine creates `$LAST30DAYS_MEMORY_DIR` automatically when running — no manual setup needed.
+# One-time: check and report config status
+bash hooks/scripts/check-config.sh
+```
+
+### Option A: Daily health check cron
+
+```bash
+hermes cron create --schedule "0 9 * * *" --prompt "Run 'bash skills/research/last30days/hooks/scripts/check-config.sh' and report the result."
+```
+
+### Option B: Manual first-run detection
+
+The first-run detection is handled by the SKILL.md Step 0 setup wizard. The engine automatically detects `~/.config/last30days/.env` and runs the wizard on first invocation. No cron needed.
+
+### Option C: Background directory creation
+
+On Hermes Desktop, the agent will create `$LAST30DAYS_MEMORY_DIR` automatically when running the skill. The directory creation is handled by the engine's save logic — the hook's `mkdir -p` is a convenience, not a requirement.
+
+## What's NOT needed on Hermes
+
+- The permission check (`chmod 600`) — Windows uses ACLs, not POSIX permissions
+- The Keychain integration — Windows uses Credential Manager (handled separately)
+- The config status message — the skill reports available sources in its own output

HERMES_SETUP.md

@@ -4,7 +4,7 @@ This guide covers installing last30days on Hermes AI Agent.
 
 ## Prerequisites
 
-1. **Hermes installed** - See https://github.com/mercurial-tf/hermes
+1. **Hermes installed** - See https://github.com/nousresearch/hermes-agent
 2. **Python 3.12+** - `brew install python@3.12` or similar
 3. **yt-dlp** (optional, for YouTube) - `brew install yt-dlp`
 
@@ -106,8 +106,29 @@ hermes skills install mvanhorn/last30days-skill --force
 
 If you symlinked your working tree (developer alternative above), just `git pull` in the repo — edits propagate live, no re-install step.
 
+## Auto-Start (SessionStart hook alternative)
+
+The Claude Code SessionStart hook (`hooks/hooks.json`) does not fire on Hermes. Instead:
+
+### Directory creation
+The engine creates `$LAST30DAYS_MEMORY_DIR` automatically when running — no manual setup needed.
+
+### Config health check
+Run manually after installing/updating API keys:
+```bash
+bash hooks/scripts/check-config.sh
+```
+
+### Optional: daily cron
+```bash
+hermes cron create --schedule "0 9 * * *" \
+  --prompt "Run 'bash skills/research/last30days/hooks/scripts/check-config.sh' and report status."
+```
+
+See `HERMES_CRON_SETUP.md` for full details.
+
 ## Support
 
 - Original repo: https://github.com/mvanhorn/last30days-skill
-- Hermes: https://github.com/mercurial-tf/hermes
+- Hermes: https://github.com/nousresearch/hermes-agent
 - Issues: Please report in the original repo

README.md

@@ -346,10 +346,12 @@ Per-client wrapper scripts, custom category-peer subreddits, and the experimenta
 
 ## Open source
 
-MIT license. No tracking. No analytics. Your research stays on your machine. 1,012 tests.
+MIT license. No tracking. No analytics. Your research stays on your machine. 1,012 tests across 144 files.
 
 Built with Python 3.12+, yt-dlp, Node.js (vendored Bird client for X search), and ScrapeCreators API. v3 engine architecture by [@j-sperling](https://github.com/j-sperling).
 
+The skill definition at `skills/last30days/SKILL.md` (66KB, under all runtime limits) is the source of truth. Detailed guides live in [`references/`](skills/last30days/references/) — loaded on demand during research execution. See [`references/INDEX.md`](skills/last30days/references/INDEX.md) for navigation.
+
 See [CONTRIBUTORS.md](CONTRIBUTORS.md) for the full list of community contributors and [CHANGELOG.md](CHANGELOG.md) for version history.
 
 ## Star History

skills/last30days/.skillignore

@@ -8,3 +8,4 @@ scripts/evaluate_search_quality.py
 scripts/test_device_auth.py
 scripts/test-v1-vs-v2.sh
 scripts/verify_v3.py
+__pycache__/

skills/last30days/references/INDEX.md

@@ -0,0 +1,34 @@
+# last30days Skill: References Index
+
+> **Navigation guide** for the 8 reference files extracted from SKILL.md.
+> The core workflow is in `SKILL.md` (66KB). These references contain detailed guides
+> loaded on demand via the agent's skill loading mechanism.
+
+## Runtime Workflow (loaded per step)
+
+| Step | File | Size | When to load |
+|------|------|------|-------------|
+| Step 0 | [`setup-wizard.md`](setup-wizard.md) | 25KB | First-run only; skip if `SETUP_COMPLETE=true` |
+| Step 0.45 | [`query-quality-preflight.md`](query-quality-preflight.md) | 7KB | When topic looks like a keyword trap |
+| Step 0.5 | [`pre-flight-resolution.md`](pre-flight-resolution.md) | 11KB | For named-entity topics (handles/repos) |
+| Step 0.55 | [`pre-research-intelligence.md`](pre-research-intelligence.md) | 16KB | For named-entity topics (communities/handles) |
+| Step 0.75 | [`query-plan-generation.md`](query-plan-generation.md) | 5KB | For named-entity topics (JSON query plan) |
+| Synthesis | [`synthesis-template.md`](synthesis-template.md) | 43KB | When synthesizing engine results |
+| Optional | [`save-html-brief.md`](save-html-brief.md) | 15KB | When emitting HTML briefs |
+
+## Reference Documentation
+
+| File | Size | Content |
+|------|------|---------|
+| [`laws-and-examples.md`](laws-and-examples.md) | 22KB | LAW 1-8 detailed explanations, violation history, worked examples |
+
+## Total
+
+- **8 files, ~143KB** extracted from the original 192KB SKILL.md
+- **Core SKILL.md** is now **66KB** (under Hermes 100KB limit)
+- **Original SKILL.md** was **192KB** (2088 lines)
+
+## Hermes Note
+
+On Hermes Agent, load reference files via `skill_view(name='last30days', file_path='references/<name>')`.
+The core SKILL.md contains short summaries and pointers to each reference file.