layered-soul/skills/codebase-knowledge-graphs/references/clawdie-graphify-integration.md
Hermes & Sam 5c5df32101 Populate layered-soul: identity, memories, skills, plan (Hermes & Sam)
- SOUL.md: full agent identity, operating principles, voice
- IDENTITY.md: runtime identity, hosts, boundaries
- USER.md: operator context imported from hermes-soul
- AGENTS.md: actual operating rules, infrastructure, quick reference
- memories/curated/: 5 topics (tailscale, forgejo, agents, projects, vaultwarden)
- skills/: 9 cross-harness skills imported from hermes-soul after review
- docs/PLAN-CONFIGURE-PRIVATE-REPO.md: configuration plan
- Validate: passes clean
2026-06-14 00:21:26 +02:00

3.4 KiB

Clawdie Graphify Integration Notes

Context

Clawdie has two coupled repositories:

  • Clawdie-AI: FreeBSD-focused agent/controlplane/runtime project.
  • Clawdie-ISO: FreeBSD/XFCE operator USB image builder and publisher.

Graphify is useful here because agents repeatedly need to understand relationships across scripts, skills, docs, firstboot modules, package lists, hostd deployment, ISO publish handoff, and USB deployment.

Use Graphify as an agent navigation and impact-analysis layer, not as a required FreeBSD runtime dependency at first.

Good uses:

  • architecture questions
  • cross-repo impact analysis
  • agent onboarding
  • firstboot/build/publish dependency map
  • hostd/webroot deployment map

Avoid initially:

  • mandatory ISO boot/build dependency
  • offline USB dependency until FreeBSD wheel/package behavior is tested
  • public/admin Telegram query feature without permission gates

Clawdie-AI Integration

Add:

.graphifyignore
docs/GRAPHIFY.md
scripts/graphify-refresh.sh

Use for questions about:

  • hostd and webroot publishing
  • Telegram command surface
  • skills discovery
  • jail/controlplane relationships
  • setup/onboarding flow

Possible future admin-only commands:

/graphify_status
/graphify_query <question>

Gate them because graph results can expose filesystem paths and repo internals.

Clawdie-ISO Integration

Add the same repo-map files:

.graphifyignore
docs/GRAPHIFY.md
scripts/graphify-refresh.sh

Use for questions about:

  • build.sh stages
  • package-list impact
  • firstboot modules
  • skills/iso-publish/SKILL.md
  • Hermes USB/IMG Deployer handoff

Exclude ISO/build artifacts aggressively:

tmp/
*.img
*.img.gz
*.iso
*.sha256
html/
downloads/
packages/
node_modules/
dist/
.cache/

Cross-Repo Graph

Generate each graph:

cd Clawdie-AI
uvx --from graphifyy graphify extract . --out .

cd ../Clawdie-ISO
uvx --from graphifyy graphify extract . --out .

Merge when asking cross-boundary questions:

uvx --from graphifyy graphify merge-graphs \
  ../Clawdie-AI/graphify-out/graph.json \
  ../Clawdie-ISO/graphify-out/graph.json \
  --out graphify-out/clawdie-merged-graph.json

Example queries:

uvx --from graphifyy graphify query \
  "what connects Clawdie-AI hostd webroot publishing to Clawdie-ISO firstboot deployment?" \
  --graph graphify-out/clawdie-merged-graph.json

uvx --from graphifyy graphify query \
  "if iso-publish manifest format changes, what docs, scripts, and skills should change?" \
  --graph graphify-out/clawdie-merged-graph.json

Suggested First Commit

Non-invasive first commit in each repo:

Add Graphify repo-map integration docs

Files:

.graphifyignore
docs/GRAPHIFY.md
scripts/graphify-refresh.sh

AGENTS.md addition:

For broad architecture/codebase questions, prefer graphify query/path/explain when graphify-out/graph.json exists. Verify source files before editing.

Validation

sh -n scripts/graphify-refresh.sh
uvx --from graphifyy graphify --help
python3 -m json.tool graphify-out/graph.json >/dev/null

Smoke queries:

uvx --from graphifyy graphify query "how does hostd publish tenant sites?" --graph graphify-out/graph.json
uvx --from graphifyy graphify query "how does iso-publish hand off to Hermes USB deployer?" --graph graphify-out/graph.json

Always compare graph answers with source files before changing code.