## Summary
- Rename `indri-services-check` task to `services-check` since it checks all services (indri native, Kubernetes, HTTP endpoints), not just indri-specific ones
- Update references in CLAUDE.md, ai-assistance-guide.md, and troubleshooting.md
## Deployment and Testing
- [ ] Run `mise run services-check` to verify the task works under its new name
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/103
## Summary
- Add "Forgejo Actions Secrets" section to forgejo reference card
- Document that `ARGOCD_AUTH_TOKEN` is used by `build-blumeops.yaml` workflow
- Note that secrets are stored in 1Password but manually copied to Forgejo (no auto-sync)
- Add missing `build-blumeops.yaml` to workflows list
- Clarify distinction between server config secrets (1Password → Ansible) vs CI/CD secrets (Forgejo UI)
## Context
The forgejo-runner ArgoCD app was showing OutOfSync because a previous attempt stored `argocd_token` in the ExternalSecret. This was incorrect - the token is actually a Forgejo Actions secret, not a k8s secret. Synced the app to remove the drift and added documentation to prevent future confusion.
🤖 Generated with [Claude Code](https://claude.ai/code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/102
## Summary
- Fix env var name from `ND_SCANSCHEDULE` to `ND_SCANNER_SCHEDULE` (Navidrome uses viper config where dots become underscores)
- Use explicit `@every 1h` format for clarity
- Reorder CLAUDE.md rules to emphasize running zk-docs first
## Root Cause
Navidrome logs showed "Periodic scan is DISABLED" at startup despite the env var being set. The config key is `scanner.schedule`, which translates to `ND_SCANNER_SCHEDULE` (not `ND_SCANSCHEDULE`).
## Deployment and Testing
- [ ] Sync navidrome app: `argocd app sync navidrome`
- [ ] Verify pod restarts with new env var
- [ ] Check logs for "Scheduling scanner" message instead of "Periodic scan is DISABLED"
- [ ] Wait ~1 hour and confirm scan runs automatically
🤖 Generated with [Claude Code](https://claude.ai/code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/101
## Summary
- Remove ~/Pictures from borgmatic source directories
- Update borgmatic and backup policy documentation
- Add Sifaka-Native Data section to clarify that photos (via Immich), music (via Navidrome), and video (via Jellyfin) are stored directly on Sifaka
## Deployment and Testing
- [ ] Run `mise run provision-indri -- --tags borgmatic --check --diff` to preview changes
- [ ] Run `mise run provision-indri -- --tags borgmatic` to apply
- [ ] Verify borgmatic config no longer includes ~/Pictures
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/100
## Summary
- Add the troubleshooting how-to guide to the zk-docs script output
- Placed after how-to/index.md to group how-to content together
## Deployment and Testing
- [x] Verified script runs successfully with `mise run zk-docs`
- [x] Changelog fragment added
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/99
## Summary
- Add `doc-random` mise task that selects a random documentation card for review
- Add how-to/knowledgebase section with review-documentation guide
- Add Caddy reference card with proxy configuration details
- Fix replication tutorial sequence (tailscale-setup now links to core-services)
- Fix "BluemeOps" typo in tailscale-setup
- Clean up obsolete zk/ directory references from doc-links
## Deployment and Testing
- [x] `mise run doc-random` works and displays a random card
- [x] `mise run doc-links` passes (all wiki-links valid)
- [x] Pre-commit hooks pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/98
## Summary
- Delete `docs/zk/` directory - all useful content migrated to structured docs
- Delete `docs/README.md` - `docs/index.md` is now the documentation root
- Add `devpi` reference card and `use-pypi-proxy` how-to guide
- Add maintenance notes to `indri` reference (sleep prevention, passwordless sudo)
- Add iCloud Photos backup note to `borgmatic` reference
- Rewrite `zk-docs` mise task to prime AI context with key docs instead of legacy cards
- Update `CLAUDE.md` and `README.md` to remove zk references
- Update `exploring-the-docs` with AI context priming section
This completes the Diataxis documentation restructuring. All six phases are now done.
## Deployment and Testing
- [x] Pre-commit hooks pass (including doc-links validator)
- [ ] Build and deploy to docs.ops.eblu.me to verify rendering
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/97
## Summary
- Create `docs/explanation/` directory with index and three explanation articles
- why-gitops: Philosophy of GitOps for homelabs (memory, rollback, AI context)
- architecture: How pieces fit together (ASCII diagrams of hosts, data flow, secrets)
- security-model: Tailscale zero-trust, 1Password secrets, access control philosophy
- Update docs/index.md with How-to and Explanation section links
- Update exploring-the-docs to link Explanation section
Decision log deferred to future work.
## Deployment and Testing
- [x] Pre-commit hooks pass (including doc-links validator)
- [ ] Build and deploy to docs.ops.eblu.me to verify rendering
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/96
## Summary
- Create `docs/how-to/` directory with index and four how-to guides
- deploy-k8s-service: Quick reference for Kubernetes deployments via ArgoCD
- add-ansible-role: Adding new Ansible roles for indri services
- update-tailscale-acls: Modifying Tailscale ACL policies via Pulumi
- troubleshooting: Diagnosing and fixing common issues
- Update exploring-the-docs to include How-to section links
- Update README.md to mark Phase 4 as complete
## Deployment and Testing
- [x] Pre-commit hooks pass (including doc-links validator)
- [ ] Build and deploy to docs.ops.eblu.me to verify rendering
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/95
The towncrier config uses the type's `directory` field as the type
identifier in filenames, NOT as subdirectories. Correct format:
docs/changelog.d/<name>.<type>.md
NOT:
docs/changelog.d/<type>/<name>.md
- Move fragments to root with type suffix
- Remove empty type subdirectories
- Fix CLAUDE.md instructions
- Fix tutorial examples in contributing.md and ai-assistance-guide.md
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Fragments in subdirectories should be named `<name>.md`, not
`<name>.<type>.md` - the type is already indicated by the directory.
- Rename feature/auto-deploy-docs.feature.md → feature/auto-deploy-docs.md
- Rename misc/+container-tag-no-confirm.misc.md → misc/+container-tag-no-confirm.md
- Update CLAUDE.md with correct fragment path format
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This ensures ArgoCD sync triggers a pod rollout when the URL changes,
since ConfigMap data changes don't restart pods automatically.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add -L flag to follow redirects
- Add -f flag to fail on HTTP errors
- Use dpkg --print-architecture as fallback for TARGETARCH
- Verify binary works after download
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
## Summary
- Add `uv` and `argocd` CLI to forgejo-runner container image
- Add `workflow-bot` ArgoCD account with sync permissions (declarative via kustomize patches)
- Add `ARGOCD_AUTH_TOKEN` to forgejo-runner external secret for workflow auth
- Update build workflow to auto-deploy docs after release:
- Update configmap with new release URL
- Commit changelog and configmap changes
- Sync docs app via ArgoCD
## Deployment and Testing
Manual steps required before this can work:
1. [ ] Build and push new forgejo-runner image (v2.4.0)
2. [ ] Sync argocd app to create workflow-bot account
3. [ ] Generate token: `argocd account generate-token --account workflow-bot`
4. [ ] Store token in 1Password under "Forgejo Secrets" with field `argocd_token`
5. [ ] Sync forgejo-runner app to pick up new external secret
6. [ ] Update forgejo-runner deployment to use new image version
7. [ ] Test by running workflow manually
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/93
- Use [[reference/index | reference]] for ambiguous index.md link
- Update doc-links to validate both filenames and paths
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Convert all wiki-links from title-based to filename-based
- Update doc-links to validate against filenames
- Add doc-filenames task for duplicate filename detection
- Consolidate doc hooks into single local block in pre-commit config
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Wiki-links with display text must use spaces around the pipe:
- Good: [[target | display]]
- Bad: [[target|display]]
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
## Summary
- Convert all frontmatter titles to lowercase-hyphenated format (e.g., `grafana-alloy` instead of `Grafana Alloy`)
- Update all wiki-links to use the new slug format
- Update `doc-titles` task to validate slug format (lowercase, hyphens only)
Quartz appears to require titles without spaces for wiki-link resolution.
## Deployment and Testing
- [x] Pre-commit hooks pass (`doc-titles` and `doc-links`)
- [ ] Build docs v1.0.8 and deploy
- [ ] Verify wiki-links resolve correctly (e.g., `[[grafana-alloy]]`)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/92
## Summary
- Remove aliases from all zk cards to prevent them from capturing wiki-links
- Convert all wiki-links from `[[filename|Title]]` to `[[Title]]` format
- Replace `doc-filenames` task with `doc-titles` for duplicate title detection
- Update pre-commit hook to use `doc-titles`
Wiki-links now resolve to reference docs by their frontmatter title, which is more readable and maintainable than filename-based links.
## Deployment and Testing
- [x] Pre-commit hooks pass (including new `doc-titles` check)
- [x] Manually verified zk cards have aliases removed
- [ ] Deploy docs v1.0.7 and verify wiki-links resolve correctly
- [ ] Test links to reference docs (e.g., [[Grafana Alloy]], [[ArgoCD]])
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/91
## Summary
- Quartz's "shortest" path mode resolves wiki-links by **filename**, not frontmatter title
- Previous PR used title-based links like `[[Grafana Alloy]]` which looked for non-existent `Grafana-Alloy.md`
- Now using filename-based links like `[[alloy|Grafana Alloy]]` which correctly resolve
## Changes
- Rename zk duplicate files with `-log` suffix (e.g., `argocd.md` → `argocd-log.md`)
- Rename `reference/storage/postgresql.md` to `postgresql-storage.md`
- Convert all 175 wiki-links from `[[Title]]` to `[[filename|Title]]` format
- Rename `doc-card-titles` task to `doc-filenames` (checks filename uniqueness, not titles)
- Update pre-commit hook for renamed task
## Deployment and Testing
- [x] Pre-commit hooks pass
- [x] `mise run doc-filenames` shows no duplicate filenames
- [ ] Verify wiki-links work correctly in Quartz build
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/90
