docs: adopt wiki-hq conventions + separate agent instructions (phases 1-2)

Problem: the narrative docs lacked an enforceable style standard, and
agent-facing instruction (OIKOS/CAVEMAN/CONTRIBUTING) was interleaved with
human content at the repo root.

Change:
- Add .agents/shared/{writing-style,llm-wiki}.md — a lint-checkable prose
  standard (with an imperative-voice exception for runbooks/recipes) and the
  sources/wiki/index/log layer model.
- Move CAVEMAN.md -> .agents/shared/caveman.md,
  CONTRIBUTING.md -> .agents/shared/page-templates.md,
  OIKOS.md -> .agents/OIKOS.md; leave thin root stubs so old links resolve.
- Add .agents/domains/{knowledge,operations}/schema.md; operations schema
  codifies "plans always live in plans/".
- Repoint live references (AGENTS, README, GLOSSARY, OIKOS) and fix OIKOS.md's
  internal relative links for its new depth.

Risk: none to the operational substrate — inventory.yaml, hosts/*.yaml,
oikos/, mcp/, secrets/, bin/ untouched (verified via git status).

Verification: relative-link check across .agents/ clean; substrate churn empty.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
2026-07-06 14:22:09 +02:00
parent 14e88c7c5e
commit bb5c0717a1
14 changed files with 814 additions and 408 deletions

227
.agents/OIKOS.md Normal file
View File

@@ -0,0 +1,227 @@
# Oikos — the operating model
Oikos (Greek: *household*) is the agent operating system layered on this
repo. It is not new infrastructure: `inventory.yaml` is the kernel data
structure, the `homelab` CLI and MCP server are the syscall surface, and
this page defines the rules everything above them follows.
Read this after [AGENTS.md](../AGENTS.md). Machine-readable companions:
[oikos/ontology.yaml](../oikos/ontology.yaml) (systems model),
[oikos/policy.yaml](../oikos/policy.yaml) (risk & approval).
## The kernel loop: OODA
Every Oikos activity — scheduled probe, agent task, operator request — is
one pass through **Observe → Orient → Decide → Act**:
1. **Observe** — probes, drift detectors, and agent findings produce
**Signals** (structured records, not loose messages): pending updates,
high temperature, low disk, service down, cert expiry, stale backup,
inventory drift.
2. **Orient** — walk the ontology graph: what entity is affected, what
depends on it (blast radius), its lifecycle state, whether a runbook
matches, what the ledger says about past attempts.
3. **Decide** — the classifier scores **risk class × blast radius ×
confidence** and routes:
- **auto-act**: within autonomy policy, high confidence, contained radius
- **escalate**: operator approval via Matrix (✅/❌ reaction) or the
Oikos Console's `/approvals` page (destructive actions additionally
need a typed confirmation phrase either way)
- **queue**: informational — console + reports
The classifier can only *lower* autonomy relative to policy, never raise
it. When in doubt, escalate.
4. **Act** — execute through `homelab` commands or runbooks (never ad-hoc
SSH), then **verify** with the action's verification command, write a
**ledger** entry, resolve the Signal, and update docs in the same session.
## Primitives
| Primitive | What it is | Lives in |
|---|---|---|
| Host / Service | topology entities | `inventory.yaml` (+ generated `hosts/*.yaml`) |
| Secret | SOPS+age encrypted value, per-client recipients | `secrets/` + `.sops.yaml` |
| Runbook | executable workflow with risk class + verification | `runbooks/` (Week 2) |
| Signal | something needing attention, with lifecycle | `signals/` ledger (Week 3) |
| Change | one mutation: who, what, risk, approval, verification | `ledger/` (Week 2) |
| Approval | short-TTL signed grant for a gated action | approval engine (Week 3) |
| Incident | investigation narrative | `investigations/` |
| Plan | design doc for non-trivial work | `plans/` |
| Agent | enrolled client identity = its age pubkey | `inventory.yaml` + `.sops.yaml` |
## Risk classes (enforced, not advisory)
From [oikos/policy.yaml](../oikos/policy.yaml):
- **read_only** — status, logs, docs, inventory. Unattended.
- **reversible_low** — restart, cache clear, sync pull. Unattended + ledger.
- **config_mutation** — tracked-config edits (commit+push, never local),
deploys, upgrades, DNS/ingress changes. Operator approval.
- **destructive** — destroy, format, wipe, rotate, revoke. Approval +
typed confirmation phrase.
Lifecycle gates modify these: `provisioning` nodes are freely mutable
(nothing depends on them); `deprecated` nodes accept no new dependents;
anything touching a `destroyed` node is drift.
## The systems model
Eight domains — physical, compute, network, storage, software,
identity & access, operations, external — cover everything in the lab;
entities are connected by typed edges (`hosts`, `provides`, `mounts`,
`stores-on`, `routes-to`, `can-decrypt`, `depends-on`, `backs-up-to`, …)
defined in [oikos/ontology.yaml](../oikos/ontology.yaml). Rule of
completeness: **if it can break, be changed, or hold data, it has an
entity and edges.** Blast-radius questions ("what breaks if strong goes
down?") are graph walks, not doc archaeology.
Nodes move through an explicit lifecycle —
`planned → provisioning → active → migrating → deprecated → destroyed`
stored as `state:` in inventory (absent = active). Destroyed nodes live in
the `archaeology:` section. Each transition is a runbook checklist;
deprecation completes only when inbound edges reach zero.
Generated views: [infrastructure/topology.md](../infrastructure/topology.md)
(Mermaid, regenerated from inventory) and the live, clickable version at
`oikos.hubris.network/graph` once the Console is deployed.
## Conventions carried forward
- Inventory is the truth; live state wins over narrative docs.
- Prefer `homelab` CLI and MCP over ad-hoc SSH.
- Meaningful changes update docs in the same session.
- Secrets are decrypted locally via per-client keys; never into docs/comments.
- Tracked configs change by commit + push, not local edits.
- Netbird is the preferred mesh path for new traffic.
- Agents are terse ([caveman.md](shared/caveman.md)), verify claims, and fix
collateral drift when found.
## Build status (30-day roadmap, started 2026-07-05)
- **Week 1**: policy, ontology, service contract, archaeology, topology
generator, this brief. Shipped.
- **Week 2**: context cards, `homelab service <name> …`, change ledger,
`node relations`, runbooks. Shipped.
- **Week 3**: ops scheduler + state cache (`homelab service <name> health`
is cache-first, `--live` forces a probe), drift detectors, signal engine
(`homelab signal …`), decision classifier (`homelab decide …`), approval
engine (`homelab approval …` — shared-HMAC grants; Matrix delivery is
Hermes's existing `@dtoro:avispero` send path, not a new bot, see
`oikos/approve.py`), daily brief + weekly report (`oikos/report.py`).
Shipped, except: Prometheus is still `planned` (see
[plans/2026-07-05-oikos-prometheus-lxc.md](../plans/2026-07-05-oikos-prometheus-lxc.md)) —
trend signals (disk-full prediction, temp creep) wait on that LXC; the
scheduler's disk check today is point-in-time only, and CPU/NVMe
temperature isn't probed at all yet (no confirmed sensor path on
hubris/strong). DNS-vs-inventory and generic tracked-config-cleanliness
drift checks are also deferred (see `oikos/drift.py` docstring).
- **Week 4**: Oikos Console v0 shipped — signals landing page, service
grid + detail, node/blast-radius view, live Mermaid graph, drift view,
approvals queue (approve/deny, destructive confirmation-phrase
enforced), daily/weekly reports. Server-rendered FastAPI + Jinja2, no
SPA build chain, tested end-to-end against live production data (see
`oikos/console/`). Deploys as a third webhook on `dtoro/Homelab-Docs`
(`/opt/oikos-console`, port :9831) — see
[oikos/console/deploy/README.md](../oikos/console/deploy/README.md) for
the Caddy route and Gitea webhook registration this repo can't do for
itself. Approval grants are now single-use (a second `check_grant` call
for the same request fails even within the TTL) and already exact-bound
to request id + entity + action.
**Not shipped as originally planned:** per-agent *age-key-signed*
request authentication — age has no signing primitive (it's an
encryption-only keypair format), so "age-key-signed" wasn't
buildable as stated. The real alternative (SSH-key signing via
`ssh-keygen -Y sign`/`-Y verify`, using each host's already-provisioned
SSH key) is real and buildable, but needs SSH public keys recorded in
inventory first — not there today. Moved to the 60/90-day backlog.
Authentik step-up re-auth on the approve/deny route is documented but
needs a live Authentik instance to configure — also backlog.
Docs pass done (this file, AGENTS.md, operations/commands.md); found
and fixed two more stale references while at it (DNS section still
pointed at destroyed LXC 124/dnsmasq instead of Technitium on 107, and
a `claudio-monitor` reference that's been deprecated since 2026-06-04).
### Real drift found while building Week 3 (unresolved, needs operator action)
The drift detectors surfaced genuine, currently-true findings on first
run against production — recorded here rather than silently fixed, since
each is a `config_mutation`/`destructive`-class decision:
- `republic-laptop` has no `age_pubkey:` in `inventory.yaml`, but its real
age key is granted on nearly every shared secret in `.sops.yaml`
(`age1vf8h7...`) — the enrollment write-back to inventory never
happened. Fix: `homelab client add republic-laptop --finalize-pubkey
age1vf8h7s8mqsn2q5eadgpdupsj4mwn8zguc77d85ws3xj40sl9rgksx2rxw6`.
- `grimmory` has an `age_pubkey` in inventory but is missing from
`secrets/hello.yaml`'s recipient list — incomplete enrollment the
other direction. Fix: re-run `homelab client add grimmory
--finalize-pubkey <its key>`.
- `pve_id 131` exists live on hubris (`pct list`) with no inventory entry
— investigate before assuming it's a stale ID (see the Prometheus LXC
plan doc above, which flags this explicitly).
- Three `lifecycle-pve-id-reuse` info findings (100, 106, 107 each shared
between an active host and an archaeology entry) — expected/benign ID
reuse after destroy, no action needed.
## 60/90-day backlog
Derived from gaps observed while building the 30-day roadmap, not
guesswork. Roughly ordered by what unblocks the most:
- **Fix the oikos-console deploy webhook's signature mismatch.** Console
is live on apps (105) via a manual `deploy.sh` run, but Gitea webhook
14's deliveries all 403 with a signature mismatch for a cause not yet
found — the secret is confirmed synced correctly on both sides
(rotated once already to rule out drift). Until fixed, `git push`
doesn't auto-redeploy the console the way it does for homelab-mcp/
secrets-issuance; re-run `deploy.sh` on apps manually after changes.
See [oikos/console/deploy/README.md](../oikos/console/deploy/README.md).
- **SSH-key-signed approval requests.** Replaces the design note in
Week 4: age keys can't sign (encryption-only format), so per-agent
request authentication needs `ssh-keygen -Y sign`/`-Y verify` against
each host's existing SSH key. Blocked on a schema gap: inventory
doesn't record SSH public keys today, only ports/users. First step is
populating that field on enrollment, then wiring `oikos/approve.py` to
require and verify a signature over the request payload.
- **Authentik step-up re-auth** on the Console's `/approvals` POST route
— needs a live Authentik `PromptStage`/reauth flow scoped to that path;
not configurable without a running instance to test against.
- **Prometheus provisioning** (see
[plans/2026-07-05-oikos-prometheus-lxc.md](../plans/2026-07-05-oikos-prometheus-lxc.md))
— unblocks trend signals (disk-full prediction, temp creep) and real
sparklines in the Console; investigate the undocumented `pve_id 131`
on hubris first.
- **CPU/NVMe temperature probing** in the scheduler — needs a confirmed
sensor path on hubris and strong (lm-sensors vs vendor tool) before a
real check can be written; guessing one risks a probe that silently
never fires.
- **DNS-vs-inventory drift check** — compare Technitium zone records
against `services.*.url`/`public_host`; not implemented (`oikos/drift.py`
has no Technitium API wiring yet).
- **Generic tracked-config-cleanliness drift check** — today only caddy's
`/etc/caddy` git-checkout path is hardcoded in `oikos/drift.py`; every
other service with a `config_repo` needs its local checkout path
recorded (a `mutation_path`-style field, same gap Week 1's service
contract flagged but didn't backfill) before this generalizes.
- **Per-service policy overrides** (`oikos/policy.yaml`
`service_overrides`) — schema is ready (caddy/dns already use it);
populate more as specific services turn out to need non-default risk
classes.
- **Incident timeline generator** — stitch ledger + signal history into
a single narrative for `investigations/` entries instead of writing
them by hand.
- **Secret access audit** — who-can-decrypt-what report from
`.sops.yaml` + inventory `age_pubkey`s, extending what
`oikos/drift.py`'s SOPS check already partially does.
- **Restore drills** — exercise `backs-up-to` (once populated) by
actually restoring from a backup target on a schedule, not just
checking freshness.
- **Multi-agent delegation model** — more than one agent acting
concurrently; needs the ledger's `agent` field to carry real identity
(age pubkey, not just hostname) consistently, which it mostly does
already but hasn't been stress-tested with concurrent writers.
- **Grafana** — only if the Console's own Prometheus-backed sparklines
turn out to be insufficient once Prometheus ships.
- **"Generalize later" extraction** — the original decision was personal-
first, generalize-later (see Week 1). Once patterns stabilize, extract
a config-driven Oikos core with no `hubris.network`/`hubris`/`strong`
hardcoding, so it's installable on a different homelab.

View File

@@ -0,0 +1,48 @@
# Knowledge domain — schema
The knowledge domain is the durable, authoritative current-state documentation of the homelab: one
page per node and per cross-cutting system, synthesized from live state and evidence. It answers
"what exists and how does it work right now."
It follows the [LLM Wiki layer model](../../shared/llm-wiki.md) and the
[writing-style](../../shared/writing-style.md) and [page-templates](../../shared/page-templates.md)
rules.
## The narrative / substrate split
The knowledge wiki is **narrative**. It sits alongside a **machine-readable substrate** that it
describes but never contains. The split is load-bearing: several programs read the substrate at
fixed paths, so the wiki reorganization never moves it.
| Layer | Location | Consumed by |
|-------|----------|-------------|
| Substrate — source of truth | `inventory.yaml` (root) | MCP server, `homelab` CLI, `oikos/` scheduler/drift/relations/gen-topology |
| Substrate — generated host records | `hosts/*.yaml` (root) | `mcp/server.py` (`HOSTS_DIR`), `bin/homelab`; written by `mcp/build_host_files.py` |
| Substrate — kernel + context cards | `oikos/` (code, `oikos/cards/`, `oikos/state.json`) | MCP `explain`, scheduler |
| Narrative — synthesized wiki | `knowledge/wiki/{hosts,containers,vms,infrastructure}/` | humans, agents via MCP `get_page` / `search_docs` |
| Evidence — immutable sources | `knowledge/sources/`, `investigations/` | synthesis into wiki pages |
## Wiki pages
- **Node pages** (`knowledge/wiki/containers/<id>-<name>.md`, `.../vms/<id>-<name>.md`,
`.../hosts/<name>.md`) follow the container/host template in
[page-templates.md](../../shared/page-templates.md): opening definition, `## At a glance`,
`## Role`, service/port map, storage, auto-deploy, `## Related`, `## Changelog`.
- **Cross-cutting pages** (`knowledge/wiki/infrastructure/<topic>.md`) follow the cross-cutting
template: `## Why`, `## Components`, `## How to apply`, `## Gotchas`, `## Related`, `## Changelog`.
- Each `inventory.yaml` host entry carries a `doc_page:` field pointing at its narrative page.
Changing where a page lives means updating that field (read by `bin/homelab`).
## The two logs
- The per-page **`## Changelog`** records infrastructure changes and is machine-parsed
(`get_changelog`, the Oikos ledger). Keep the `### YYYY-MM-DD — title` shape.
- **`knowledge/log.md`** is append-only and records *documentation-maintenance* operations only
(restructures, source ingests, lint sweeps): `## [YYYY-MM-DD] <op> | <summary>`. It never
duplicates the Oikos change ledger (`oikos/ledger.py`).
## Same-session update rule
A change to a node updates every page that references it in the same session — the node page, the
section `README.md` table, the root `README.md`, the Caddy/DNS/ingress pages, the host page, and
`inventory.yaml`. See [page-templates.md](../../shared/page-templates.md#same-session-update-rule).

View File

@@ -0,0 +1,50 @@
# Operations domain — schema
The operations domain holds the procedural and time-stamped documentation: runbooks (repeatable
procedures), investigations (incident evidence), and plans (design docs for non-trivial work). It
follows [writing-style](../../shared/writing-style.md); runbooks and plans use the imperative voice
exception.
## Plans always live in `plans/`
**Any plan or design doc for the Homelab is written into the repo `plans/` folder as
`plans/YYYY-MM-DD-slug.md` — never a scratch path, an agent-private plan location, or a chat
message.** An agent drafting a plan:
1. Writes the file under `plans/` using the plan template in [page-templates.md](../../shared/page-templates.md).
2. Lists it in `plans/index.md`.
3. On completion, moves it to `plans/done/` and updates the index status.
This is the single source for homelab design intent; keeping it in-repo means the plan is
versioned, reviewable, and reachable by MCP `get_page`/`search_docs` like any other doc.
## Runbooks
Repeatable procedures live in `runbooks/<name>.md` with YAML front-matter that the Oikos policy and
lifecycle machinery reads:
```yaml
---
name: <name>
risk_class: read_only | reversible_low | config_mutation | destructive
inputs: [<param>, ...]
verification: "<shell expression that proves success>"
docs_update_checklist: [<doc artifacts to update>]
transition: "<from> -> <to>" # only for lifecycle runbooks
---
```
`risk_class` values and the lifecycle `transition` states must match
[`oikos/policy.yaml`](../../../oikos/policy.yaml) and [`oikos/ontology.yaml`](../../../oikos/ontology.yaml).
## Investigations
Incident records live in `investigations/YYYY-MM-DD-slug.md` and are **evidence sources** — written
once at incident time, then linked from the changelogs of the nodes they implicate. Sections:
`## Summary`, `## Timeline`, `## Root cause`, `## Mitigations applied`, `## Open questions`. Resolved
incidents move to `investigations/archive/`.
## The operations log
`plans/log.md` and `investigations/log.md` are append-only records of documentation operations on
those areas (`## [YYYY-MM-DD] <op> | <summary>`), distinct from the Oikos change ledger.

33
.agents/shared/caveman.md Normal file
View File

@@ -0,0 +1,33 @@
# CAVEMAN.md — communication mode for homelab agents
Respond terse like smart caveman. All technical substance stay. Only fluff die.
## Rules
Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted exact.
Pattern: `[thing] [action] [reason]. [next step].`
Not: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..."
Yes: "Bug in auth middleware. Token expiry check use `<` not `<=`. Fix:"
## Levels
- **lite** — no filler/hedging. Keep articles + full sentences. Professional but tight.
- **full** (default) — drop articles, fragments OK, short synonyms. Classic caveman.
- **ultra** — abbreviate prose words (DB/auth/config/req/res), strip conjunctions, arrows (X → Y). Code symbols/API names/errors: never abbreviate.
Switch: `/caveman lite|full|ultra`. Stop: "normal mode".
## Auto-Clarity
Drop caveman for: security warnings, irreversible actions, multi-step sequences where fragments risk misread, user confused/repeating. Resume after clear part done.
## Boundaries
Code/commits/PRs: write normal. "stop caveman" or "normal mode": revert. Level persist until changed or session end.
---
Source: https://github.com/JuliusBrussee/caveman
Copy to `~/.hermes/skills/` for Hermes Agent, or `~/.claude/projects/<name>/SKILL.md` for Claude Code.

View File

@@ -0,0 +1,41 @@
# LLM Wiki — the documentation contract
How the narrative documentation in this repo is organized. The pattern is borrowed from the
`sources / wiki / index / log` model: a durable synthesized layer (`knowledge/wiki/`) built on top
of immutable evidence (`knowledge/sources/`, `investigations/`), with pure-listing indexes and an
append-only operations log.
This contract governs the **narrative layer only**. The machine-readable substrate — `inventory.yaml`,
generated `hosts/*.yaml`, `oikos/`, `mcp/`, `secrets/`, `bin/` — is not part of the wiki and never
moves under it. See [the knowledge schema](../domains/knowledge/schema.md) for the split.
## Layers
- **Sources** are immutable raw material: incident records (`investigations/`), external reference
docs (`knowledge/sources/references/`), and the live system itself (`pct config`, `docker inspect`).
Read them; do not rewrite them into other sources.
- **Wiki** (`knowledge/wiki/`) is the synthesized, authoritative current-state layer: one page per
node (`containers/`, `vms/`, host narratives) and per cross-cutting system (`infrastructure/`). A
reader understands the topic from the wiki page without reading the sources.
- **Index** (`index.md` / folder `README.md`) is a pure listing — every page in scope with a
one-line summary, and nothing else. Anything the section wants to say up front goes into a page
the index lists, not into the index.
- **Log** (`log.md`) is append-only, recording *doc-maintenance operations* (restructures, source
ingests, lint sweeps) in single-line format: `## [YYYY-MM-DD] <op> | <summary>`.
## Two logs, kept distinct
- **`## Changelog`** on each node/topic page records *infrastructure* changes to that node. It is
machine-parsed (`get_changelog`, the Oikos ledger) — keep the `### YYYY-MM-DD — title` shape.
- **`log.md`** per area records *documentation* operations only. It never duplicates the Oikos
change ledger (`oikos/ledger.py`), which stays authoritative for infra changes with
who/what/risk/approval/verification.
## Rules
- Wiki pages stay short and focused. A page past ~300 lines splits.
- Pages stay flat under `wiki/<section>/` until there are enough to warrant a sub-group.
- Every page follows [writing-style.md](writing-style.md).
- Plans and design docs always live in the repo `plans/` folder (`plans/YYYY-MM-DD-slug.md`),
listed in `plans/index.md`, moved to `plans/done/` on completion — never a scratch path or a chat
message. See [the operations schema](../domains/operations/schema.md).

View File

@@ -0,0 +1,152 @@
# Page templates for the Homelab Wiki
The structural templates for each page type. Prose voice, vocabulary, and cross-reference rules live
in [writing-style.md](writing-style.md); the layer model (sources / wiki / index / log) lives in
[llm-wiki.md](llm-wiki.md).
## Voice
Concise, technical, sysadmin-to-sysadmin. No marketing prose, no exclamation marks. Full rules in
[writing-style.md](writing-style.md).
## Page templates
### Container page (`containers/<id>-<name>.md`)
```markdown
# <id> — `<name>`
One-sentence purpose.
## At a glance
- **Hostname:** `<name>`
- **IP:** `192.168.8.x`
- **Privilege:** privileged | unprivileged
- **Resources:** N cores / M GiB RAM / D GiB rootfs
- **Mounts:** `/mnt/library``/mnt/library` (if any)
- **Public hostname:** `<sub>.hubris.network` (if proxied)
## Role
What it does, what it talks to.
## Service / port map
| Service | Listen | Notes |
## Storage / config paths
## Auto-deploy
(if any) — link to [auto-deploy](../infrastructure/auto-deploy.md)
## Related
- [Caddy](121-caddy.md) (if proxied)
- [DNS](../infrastructure/dns.md) (if has subdomain)
- [Authentik](124-authentik.md) (if SSO)
- ...
## Changelog
### YYYY-MM-DD — short title
What changed, why, link to investigation if any.
```
### Cross-cutting page (`infrastructure/<topic>.md`)
```markdown
# <Topic>
One-sentence summary.
## Why
Design rationale — what it replaces, what it solves.
## Components
Where it runs, what files matter.
## How to apply / use
Recipes.
## Gotchas
## Related
Links to nodes that host or depend on this.
## Changelog
```
### Plan (`plans/YYYY-MM-DD-slug.md`)
```markdown
# YYYY-MM-DD — <title>
## Goal
What this change achieves and why.
## Current topology / state
Diagram or description of what exists now.
## Target topology / state
What it looks like after.
## Pre-flight checklist
## Step-by-step procedure
## Verification
## Post-migration
Changelog entries to write, index status to update.
```
### Investigation (`investigations/YYYY-MM-DD-slug.md`)
```markdown
# YYYY-MM-DD — <title>
## Summary
1-3 sentences.
## Timeline
## Root cause
## Mitigations applied
## Open questions
```
## Linking discipline
- Every container page links to every cross-cutting page it participates in.
- Every cross-cutting page lists the nodes that participate.
- Every investigation links to the nodes it implicates *and* gets back-linked from each node's changelog.
- Every plan links to the infrastructure pages it affects. When done, update the plan's status in `plans/index.md` and write changelog entries on affected node pages.
## Changelog hygiene
- Reverse-chronological (newest first).
- One entry per discrete change, even if you make several in one day.
- If a change spans nodes, repeat the entry on each affected page (different perspective is fine).
- Don't rewrite history — entries are append-only. Mistakes get a follow-up entry that supersedes them.
## Same-session update rule
When you make a change to a node — migrate an LXC, update an IP, change a
mount, deploy a new service — **update every relevant doc page in the same
session.** A change that touches a container page must also update:
- The `containers/index.md` table (IPs, host, mounts, status)
- The `README.md` table (if the change affects listed columns)
- The Caddy page site list (if the change affects `*.hubris.network` routing)
- The DNS / ingress infrastructure pages (if the change affects routing)
- The `hosts/{hubris,strong}.md` host page (if container count changes)
- The `inventory.yaml` host entry (source of truth for the `hosts/*.yaml` generation)
- The `infrastructure/topology.md` (generated from inventory, but regen if needed)
The pattern of updating only one page and leaving stale references on others
is a bug. If you're doing a multi-step migration, document the intermediate
state with a changelog entry that says "pending — will finalize after Phase
N."
This rule is why Phase 2 of the strong migration (2026-07-05) caused
widespread stale data: individual container pages were updated in the
changelog but never had their At-a-glance sections, IPs, mount paths, or
host attribution updated. Don't repeat that.

View File

@@ -0,0 +1,75 @@
# Writing Style
Write like a technical reference, not a marketing page. Every sentence conveys new information.
These rules govern **committed documentation** — wiki pages, READMEs, schemas, skills, `AGENTS.md`,
plans, investigations, and code comments. They are separate from [caveman.md](caveman.md), which
governs an agent's *chat responses*; the two do not conflict.
New or rewritten pages follow these patterns from day one. Existing pages get updated the next time
they are touched.
## Vocabulary — never use these
- Significance puffers: "pivotal", "crucial", "vital", "groundbreaking", "transformative", "testament", "paramount", "invaluable".
- Analytical verbs: "delve", "leverage", "utilize", "facilitate", "foster", "showcase", "underscore", "streamline", "harness".
- Poetic nouns: "tapestry", "landscape" (figurative), "realm", "paradigm", "ecosystem" (figurative), "journey" (figurative), "nexus", "cornerstone".
- Promotional adjectives: "robust", "seamless", "innovative", "cutting-edge", "meticulous", "holistic", "comprehensive".
- Opening crutches: "In today's world", "In the ever-evolving landscape of", "It's worth noting that", "It is important to note that".
Use short, common words: "use" not "utilize", "help" not "facilitate", "show" not "demonstrate".
## Voice
Describe what systems do and how they work.
- **Reference prose** (node pages, cross-cutting infrastructure descriptions, `## Role`, `## Why`,
`At a glance`) is third-person: state facts about the system, not instructions to a reader.
- **Recipes, runbooks, and skills** are the exception: second-person imperative is allowed and
preferred where it makes a procedure clearer ("Edit the Caddyfile, commit + push", "Verify with
`dig +short`"). This matches how the operator actually works. The vocabulary, structure, and
cross-reference rules below still apply.
## Page shape
Every doc-level page follows the same shape so a reader scans it in one pass.
1. **One H1 = the page title.** Node pages use `# <id> — \`<name>\``; topic pages use `# <Topic>`.
2. **Opening definition.** First paragraph, 13 sentences, says what the thing is. No motivation, no marketing, no setup.
3. **Body sections** in the natural order for the topic. Reuse the section templates in [page-templates.md](page-templates.md).
4. **`## Changelog`** at the bottom of every node/topic page — reverse-chronological, append-only. This section is machine-parsed (`get_changelog` in `mcp/server.py`); keep the `### YYYY-MM-DD — title` shape.
5. **Related links** only at the bottom, only when a reference cannot be woven inline.
## Section indexes (folder READMEs)
A folder's `README.md` opens with a 13 sentence prose intro that says what the section covers, then
a single navigation table — `| Document | What it covers |` — and nothing else. No stale counts, no
duplicated prose, no narrative between the intro and the table.
## Structure rules
- Make every sentence information-dense. Cut filler, qualifiers, and setup phrases. Lead with the concrete fact or action, not why it matters.
- No participial tack-ons (", highlighting the importance of…"). If the clause adds information, make it a separate sentence.
- **No meta-commentary about the content itself.** Do not narrate the page's own structure or linking strategy.
- Prefer **tables** for enumerable items with internal structure (service/port maps, field lists, status grids). Reserve bullets for short non-structured lists.
- Use the **bold-leading-phrase pattern** for structured points: `**Read-only by construction.** The MCP server never mutates state.` — a bold noun phrase, a period, then the explanation.
- When enumerating across services or nodes, give each its own `###` sub-section or a table row, not one run-on paragraph.
- Use backticks for code, paths, hostnames, and file names (`inventory.yaml`, `192.168.8.77`, `pct config`); italics for first-mention terminology.
- Use `>` blockquotes for caveats and gaps that interrupt the main flow: `> **Outstanding gap.** DNS-vs-inventory drift check not yet wired.` One thought per blockquote.
## Diagrams
- Mermaid is the default for topology and flow diagrams. `infrastructure/topology.md` is generated by `oikos/gen-topology.py` — do not hand-edit it.
- ASCII box diagrams are fine for small shape diagrams; keep them to one screen.
## Sourcing and cross-references
- **Factual discipline.** Every claim is grounded in a cited source, an adjacent linked page, or a directly observable fact (`pct config`, `docker inspect`, running config). Do not write sentences that sound sourced but are inference. When docs disagree with live state, fix the doc and note it in the changelog.
- **One-sided cross-references.** When two pages relate, the link lives in the page where the connection makes organizational sense. Do not add a back-pointer unless that direction also carries content the reader needs.
- **Cross-references are content, not catalog.** Inline links arise from the surrounding prose; the linked page must be needed to understand the current sentence. A bottom-of-page "Related" list is the fallback, not the default.
- Pages link with standard relative markdown links (e.g. a container page links to `../infrastructure/dns.md`), forming a navigable graph. Orphans are a bug.
## Code comments and commit/PR prose
- Comments explain intent, trade-offs, or constraints the code cannot convey. No diff narration, no type restatement, no section-divider comments.
- Commit messages and PR descriptions are problem → change → risk → verification, not a file-by-file diff restatement.
- The banned vocabulary applies the same way in comments and commit messages.

View File

@@ -5,10 +5,19 @@ context is in this checkout at `/opt/homelab-context/`. This file is the entry
point. Read it once at start, then keep working.
The operating model — OODA loop, risk classes, approval rules, the ontology,
and node lifecycle — is defined in [OIKOS.md](OIKOS.md). Before any mutation,
and node lifecycle — is defined in [OIKOS.md](.agents/OIKOS.md). Before any mutation,
classify the action against `oikos/policy.yaml`; when the class requires
approval, stop and ask the operator.
Agent-facing instruction is separated from human content under `.agents/`:
`.agents/shared/` holds the conventions every agent applies
([writing-style](.agents/shared/writing-style.md), [caveman](.agents/shared/caveman.md),
[page-templates](.agents/shared/page-templates.md), [llm-wiki](.agents/shared/llm-wiki.md)), and
`.agents/domains/` holds the per-domain schemas
([knowledge](.agents/domains/knowledge/schema.md), [operations](.agents/domains/operations/schema.md)).
The narrative wiki lives under `knowledge/wiki/`; the machine-readable substrate
(`inventory.yaml`, `hosts/*.yaml`, `oikos/`) stays at the repo root.
## 1. Who you are
Run `hostname` (Linux) or `scutil --get LocalHostName` (macOS), then read:
@@ -102,7 +111,7 @@ Grep is fine for browsing or when MCP is unreachable.
## 6. Communication mode
Read and apply `/opt/homelab-context/CAVEMAN.md` (if present). It defines the lab's
Read and apply `/opt/homelab-context/.agents/shared/caveman.md` (if present). It defines the lab's
terse-communication standard — drop filler, keep substance, use fragments.
## 7. Auto-setup mechanism

View File

@@ -1,33 +1,3 @@
# CAVEMAN.md — communication mode for homelab agents
# CAVEMAN.md — moved
Respond terse like smart caveman. All technical substance stay. Only fluff die.
## Rules
Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted exact.
Pattern: `[thing] [action] [reason]. [next step].`
Not: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..."
Yes: "Bug in auth middleware. Token expiry check use `<` not `<=`. Fix:"
## Levels
- **lite** — no filler/hedging. Keep articles + full sentences. Professional but tight.
- **full** (default) — drop articles, fragments OK, short synonyms. Classic caveman.
- **ultra** — abbreviate prose words (DB/auth/config/req/res), strip conjunctions, arrows (X → Y). Code symbols/API names/errors: never abbreviate.
Switch: `/caveman lite|full|ultra`. Stop: "normal mode".
## Auto-Clarity
Drop caveman for: security warnings, irreversible actions, multi-step sequences where fragments risk misread, user confused/repeating. Resume after clear part done.
## Boundaries
Code/commits/PRs: write normal. "stop caveman" or "normal mode": revert. Level persist until changed or session end.
---
Source: https://github.com/JuliusBrussee/caveman
Copy to `~/.hermes/skills/` for Hermes Agent, or `~/.claude/projects/<name>/SKILL.md` for Claude Code.
Agent chat-mode rules now live at [`.agents/shared/caveman.md`](.agents/shared/caveman.md).

View File

@@ -1,147 +1,7 @@
# Contributing to the Homelab Wiki
# CONTRIBUTING.md — moved
## Voice
Documentation conventions are now split across `.agents/shared/`:
Concise, technical, sysadmin-to-sysadmin. No marketing prose, no exclamation marks.
## Page templates
### Container page (`containers/<id>-<name>.md`)
```markdown
# <id> — `<name>`
One-sentence purpose.
## At a glance
- **Hostname:** `<name>`
- **IP:** `192.168.8.x`
- **Privilege:** privileged | unprivileged
- **Resources:** N cores / M GiB RAM / D GiB rootfs
- **Mounts:** `/mnt/library``/mnt/library` (if any)
- **Public hostname:** `<sub>.hubris.network` (if proxied)
## Role
What it does, what it talks to.
## Service / port map
| Service | Listen | Notes |
## Storage / config paths
## Auto-deploy
(if any) — link to [auto-deploy](../infrastructure/auto-deploy.md)
## Related
- [Caddy](121-caddy.md) (if proxied)
- [DNS](../infrastructure/dns.md) (if has subdomain)
- [Authentik](124-authentik.md) (if SSO)
- ...
## Changelog
### YYYY-MM-DD — short title
What changed, why, link to investigation if any.
```
### Cross-cutting page (`infrastructure/<topic>.md`)
```markdown
# <Topic>
One-sentence summary.
## Why
Design rationale — what it replaces, what it solves.
## Components
Where it runs, what files matter.
## How to apply / use
Recipes.
## Gotchas
## Related
Links to nodes that host or depend on this.
## Changelog
```
### Plan (`plans/YYYY-MM-DD-slug.md`)
```markdown
# YYYY-MM-DD — <title>
## Goal
What this change achieves and why.
## Current topology / state
Diagram or description of what exists now.
## Target topology / state
What it looks like after.
## Pre-flight checklist
## Step-by-step procedure
## Verification
## Post-migration
Changelog entries to write, index status to update.
```
### Investigation (`investigations/YYYY-MM-DD-slug.md`)
```markdown
# YYYY-MM-DD — <title>
## Summary
1-3 sentences.
## Timeline
## Root cause
## Mitigations applied
## Open questions
```
## Linking discipline
- Every container page links to every cross-cutting page it participates in.
- Every cross-cutting page lists the nodes that participate.
- Every investigation links to the nodes it implicates *and* gets back-linked from each node's changelog.
- Every plan links to the infrastructure pages it affects. When done, update the plan's status in `plans/index.md` and write changelog entries on affected node pages.
## Changelog hygiene
- Reverse-chronological (newest first).
- One entry per discrete change, even if you make several in one day.
- If a change spans nodes, repeat the entry on each affected page (different perspective is fine).
- Don't rewrite history — entries are append-only. Mistakes get a follow-up entry that supersedes them.
## Same-session update rule
When you make a change to a node — migrate an LXC, update an IP, change a
mount, deploy a new service — **update every relevant doc page in the same
session.** A change that touches a container page must also update:
- The `containers/index.md` table (IPs, host, mounts, status)
- The `README.md` table (if the change affects listed columns)
- The Caddy page site list (if the change affects `*.hubris.network` routing)
- The DNS / ingress infrastructure pages (if the change affects routing)
- The `hosts/{hubris,strong}.md` host page (if container count changes)
- The `inventory.yaml` host entry (source of truth for the `hosts/*.yaml` generation)
- The `infrastructure/topology.md` (generated from inventory, but regen if needed)
The pattern of updating only one page and leaving stale references on others
is a bug. If you're doing a multi-step migration, document the intermediate
state with a changelog entry that says "pending — will finalize after Phase
N."
This rule is why Phase 2 of the strong migration (2026-07-05) caused
widespread stale data: individual container pages were updated in the
changelog but never had their At-a-glance sections, IPs, mount paths, or
host attribution updated. Don't repeat that.
- [`page-templates.md`](.agents/shared/page-templates.md) — per-page-type structural templates, linking discipline, changelog hygiene, same-session update rule.
- [`writing-style.md`](.agents/shared/writing-style.md) — prose voice, banned vocabulary, page shape.
- [`llm-wiki.md`](.agents/shared/llm-wiki.md) — the sources / wiki / index / log layer model.

View File

@@ -18,7 +18,7 @@ Terms and abbreviations used throughout the homelab wiki.
| **Mesh** | Overlay VPN for off-LAN connectivity. Netbird is current; Tailscale is legacy |
| **Netbird** | Preferred mesh VPN. VPS hosts the management plane; all homelab nodes are members |
| **OIDC** | OpenID Connect. Protocol used by Authentik for SSO login flows |
| **Oikos** | Agent operating model (OIKOS.md). OODA loop, risk classes, policy, ontology |
| **Oikos** | Agent operating model ([.agents/OIKOS.md](.agents/OIKOS.md)). OODA loop, risk classes, policy, ontology |
| **PVE** | Proxmox Virtual Environment — the hypervisor on both hubris and strong |
| **SOPS** | `sops` — Mozilla SOPS. Encrypts secrets with age keys so they live in the git repo |
| **Strong** | Secondary Proxmox VE node. Cluster member 2 (hostname `strong`, nickname ludo/ludo-mini) |
@@ -30,4 +30,4 @@ Terms and abbreviations used throughout the homelab wiki.
## See also
- [Infrastructure index](infrastructure/index.md) — cross-cutting systems each with their own doc page
- [OIKOS operating model](OIKOS.md) — agent policy, risk classes, lifecycle
- [OIKOS operating model](.agents/OIKOS.md) — agent policy, risk classes, lifecycle

228
OIKOS.md
View File

@@ -1,227 +1,3 @@
# Oikos — the operating model
# OIKOS.md — moved
Oikos (Greek: *household*) is the agent operating system layered on this
repo. It is not new infrastructure: `inventory.yaml` is the kernel data
structure, the `homelab` CLI and MCP server are the syscall surface, and
this page defines the rules everything above them follows.
Read this after [AGENTS.md](AGENTS.md). Machine-readable companions:
[oikos/ontology.yaml](oikos/ontology.yaml) (systems model),
[oikos/policy.yaml](oikos/policy.yaml) (risk & approval).
## The kernel loop: OODA
Every Oikos activity — scheduled probe, agent task, operator request — is
one pass through **Observe → Orient → Decide → Act**:
1. **Observe** — probes, drift detectors, and agent findings produce
**Signals** (structured records, not loose messages): pending updates,
high temperature, low disk, service down, cert expiry, stale backup,
inventory drift.
2. **Orient** — walk the ontology graph: what entity is affected, what
depends on it (blast radius), its lifecycle state, whether a runbook
matches, what the ledger says about past attempts.
3. **Decide** — the classifier scores **risk class × blast radius ×
confidence** and routes:
- **auto-act**: within autonomy policy, high confidence, contained radius
- **escalate**: operator approval via Matrix (✅/❌ reaction) or the
Oikos Console's `/approvals` page (destructive actions additionally
need a typed confirmation phrase either way)
- **queue**: informational — console + reports
The classifier can only *lower* autonomy relative to policy, never raise
it. When in doubt, escalate.
4. **Act** — execute through `homelab` commands or runbooks (never ad-hoc
SSH), then **verify** with the action's verification command, write a
**ledger** entry, resolve the Signal, and update docs in the same session.
## Primitives
| Primitive | What it is | Lives in |
|---|---|---|
| Host / Service | topology entities | `inventory.yaml` (+ generated `hosts/*.yaml`) |
| Secret | SOPS+age encrypted value, per-client recipients | `secrets/` + `.sops.yaml` |
| Runbook | executable workflow with risk class + verification | `runbooks/` (Week 2) |
| Signal | something needing attention, with lifecycle | `signals/` ledger (Week 3) |
| Change | one mutation: who, what, risk, approval, verification | `ledger/` (Week 2) |
| Approval | short-TTL signed grant for a gated action | approval engine (Week 3) |
| Incident | investigation narrative | `investigations/` |
| Plan | design doc for non-trivial work | `plans/` |
| Agent | enrolled client identity = its age pubkey | `inventory.yaml` + `.sops.yaml` |
## Risk classes (enforced, not advisory)
From [oikos/policy.yaml](oikos/policy.yaml):
- **read_only** — status, logs, docs, inventory. Unattended.
- **reversible_low** — restart, cache clear, sync pull. Unattended + ledger.
- **config_mutation** — tracked-config edits (commit+push, never local),
deploys, upgrades, DNS/ingress changes. Operator approval.
- **destructive** — destroy, format, wipe, rotate, revoke. Approval +
typed confirmation phrase.
Lifecycle gates modify these: `provisioning` nodes are freely mutable
(nothing depends on them); `deprecated` nodes accept no new dependents;
anything touching a `destroyed` node is drift.
## The systems model
Eight domains — physical, compute, network, storage, software,
identity & access, operations, external — cover everything in the lab;
entities are connected by typed edges (`hosts`, `provides`, `mounts`,
`stores-on`, `routes-to`, `can-decrypt`, `depends-on`, `backs-up-to`, …)
defined in [oikos/ontology.yaml](oikos/ontology.yaml). Rule of
completeness: **if it can break, be changed, or hold data, it has an
entity and edges.** Blast-radius questions ("what breaks if strong goes
down?") are graph walks, not doc archaeology.
Nodes move through an explicit lifecycle —
`planned → provisioning → active → migrating → deprecated → destroyed`
stored as `state:` in inventory (absent = active). Destroyed nodes live in
the `archaeology:` section. Each transition is a runbook checklist;
deprecation completes only when inbound edges reach zero.
Generated views: [infrastructure/topology.md](infrastructure/topology.md)
(Mermaid, regenerated from inventory) and the live, clickable version at
`oikos.hubris.network/graph` once the Console is deployed.
## Conventions carried forward
- Inventory is the truth; live state wins over narrative docs.
- Prefer `homelab` CLI and MCP over ad-hoc SSH.
- Meaningful changes update docs in the same session.
- Secrets are decrypted locally via per-client keys; never into docs/comments.
- Tracked configs change by commit + push, not local edits.
- Netbird is the preferred mesh path for new traffic.
- Agents are terse ([CAVEMAN.md](CAVEMAN.md)), verify claims, and fix
collateral drift when found.
## Build status (30-day roadmap, started 2026-07-05)
- **Week 1**: policy, ontology, service contract, archaeology, topology
generator, this brief. Shipped.
- **Week 2**: context cards, `homelab service <name> …`, change ledger,
`node relations`, runbooks. Shipped.
- **Week 3**: ops scheduler + state cache (`homelab service <name> health`
is cache-first, `--live` forces a probe), drift detectors, signal engine
(`homelab signal …`), decision classifier (`homelab decide …`), approval
engine (`homelab approval …` — shared-HMAC grants; Matrix delivery is
Hermes's existing `@dtoro:avispero` send path, not a new bot, see
`oikos/approve.py`), daily brief + weekly report (`oikos/report.py`).
Shipped, except: Prometheus is still `planned` (see
[plans/2026-07-05-oikos-prometheus-lxc.md](plans/2026-07-05-oikos-prometheus-lxc.md)) —
trend signals (disk-full prediction, temp creep) wait on that LXC; the
scheduler's disk check today is point-in-time only, and CPU/NVMe
temperature isn't probed at all yet (no confirmed sensor path on
hubris/strong). DNS-vs-inventory and generic tracked-config-cleanliness
drift checks are also deferred (see `oikos/drift.py` docstring).
- **Week 4**: Oikos Console v0 shipped — signals landing page, service
grid + detail, node/blast-radius view, live Mermaid graph, drift view,
approvals queue (approve/deny, destructive confirmation-phrase
enforced), daily/weekly reports. Server-rendered FastAPI + Jinja2, no
SPA build chain, tested end-to-end against live production data (see
`oikos/console/`). Deploys as a third webhook on `dtoro/Homelab-Docs`
(`/opt/oikos-console`, port :9831) — see
[oikos/console/deploy/README.md](oikos/console/deploy/README.md) for
the Caddy route and Gitea webhook registration this repo can't do for
itself. Approval grants are now single-use (a second `check_grant` call
for the same request fails even within the TTL) and already exact-bound
to request id + entity + action.
**Not shipped as originally planned:** per-agent *age-key-signed*
request authentication — age has no signing primitive (it's an
encryption-only keypair format), so "age-key-signed" wasn't
buildable as stated. The real alternative (SSH-key signing via
`ssh-keygen -Y sign`/`-Y verify`, using each host's already-provisioned
SSH key) is real and buildable, but needs SSH public keys recorded in
inventory first — not there today. Moved to the 60/90-day backlog.
Authentik step-up re-auth on the approve/deny route is documented but
needs a live Authentik instance to configure — also backlog.
Docs pass done (this file, AGENTS.md, operations/commands.md); found
and fixed two more stale references while at it (DNS section still
pointed at destroyed LXC 124/dnsmasq instead of Technitium on 107, and
a `claudio-monitor` reference that's been deprecated since 2026-06-04).
### Real drift found while building Week 3 (unresolved, needs operator action)
The drift detectors surfaced genuine, currently-true findings on first
run against production — recorded here rather than silently fixed, since
each is a `config_mutation`/`destructive`-class decision:
- `republic-laptop` has no `age_pubkey:` in `inventory.yaml`, but its real
age key is granted on nearly every shared secret in `.sops.yaml`
(`age1vf8h7...`) — the enrollment write-back to inventory never
happened. Fix: `homelab client add republic-laptop --finalize-pubkey
age1vf8h7s8mqsn2q5eadgpdupsj4mwn8zguc77d85ws3xj40sl9rgksx2rxw6`.
- `grimmory` has an `age_pubkey` in inventory but is missing from
`secrets/hello.yaml`'s recipient list — incomplete enrollment the
other direction. Fix: re-run `homelab client add grimmory
--finalize-pubkey <its key>`.
- `pve_id 131` exists live on hubris (`pct list`) with no inventory entry
— investigate before assuming it's a stale ID (see the Prometheus LXC
plan doc above, which flags this explicitly).
- Three `lifecycle-pve-id-reuse` info findings (100, 106, 107 each shared
between an active host and an archaeology entry) — expected/benign ID
reuse after destroy, no action needed.
## 60/90-day backlog
Derived from gaps observed while building the 30-day roadmap, not
guesswork. Roughly ordered by what unblocks the most:
- **Fix the oikos-console deploy webhook's signature mismatch.** Console
is live on apps (105) via a manual `deploy.sh` run, but Gitea webhook
14's deliveries all 403 with a signature mismatch for a cause not yet
found — the secret is confirmed synced correctly on both sides
(rotated once already to rule out drift). Until fixed, `git push`
doesn't auto-redeploy the console the way it does for homelab-mcp/
secrets-issuance; re-run `deploy.sh` on apps manually after changes.
See [oikos/console/deploy/README.md](oikos/console/deploy/README.md).
- **SSH-key-signed approval requests.** Replaces the design note in
Week 4: age keys can't sign (encryption-only format), so per-agent
request authentication needs `ssh-keygen -Y sign`/`-Y verify` against
each host's existing SSH key. Blocked on a schema gap: inventory
doesn't record SSH public keys today, only ports/users. First step is
populating that field on enrollment, then wiring `oikos/approve.py` to
require and verify a signature over the request payload.
- **Authentik step-up re-auth** on the Console's `/approvals` POST route
— needs a live Authentik `PromptStage`/reauth flow scoped to that path;
not configurable without a running instance to test against.
- **Prometheus provisioning** (see
[plans/2026-07-05-oikos-prometheus-lxc.md](plans/2026-07-05-oikos-prometheus-lxc.md))
— unblocks trend signals (disk-full prediction, temp creep) and real
sparklines in the Console; investigate the undocumented `pve_id 131`
on hubris first.
- **CPU/NVMe temperature probing** in the scheduler — needs a confirmed
sensor path on hubris and strong (lm-sensors vs vendor tool) before a
real check can be written; guessing one risks a probe that silently
never fires.
- **DNS-vs-inventory drift check** — compare Technitium zone records
against `services.*.url`/`public_host`; not implemented (`oikos/drift.py`
has no Technitium API wiring yet).
- **Generic tracked-config-cleanliness drift check** — today only caddy's
`/etc/caddy` git-checkout path is hardcoded in `oikos/drift.py`; every
other service with a `config_repo` needs its local checkout path
recorded (a `mutation_path`-style field, same gap Week 1's service
contract flagged but didn't backfill) before this generalizes.
- **Per-service policy overrides** (`oikos/policy.yaml`
`service_overrides`) — schema is ready (caddy/dns already use it);
populate more as specific services turn out to need non-default risk
classes.
- **Incident timeline generator** — stitch ledger + signal history into
a single narrative for `investigations/` entries instead of writing
them by hand.
- **Secret access audit** — who-can-decrypt-what report from
`.sops.yaml` + inventory `age_pubkey`s, extending what
`oikos/drift.py`'s SOPS check already partially does.
- **Restore drills** — exercise `backs-up-to` (once populated) by
actually restoring from a backup target on a schedule, not just
checking freshness.
- **Multi-agent delegation model** — more than one agent acting
concurrently; needs the ledger's `agent` field to carry real identity
(age pubkey, not just hostname) consistently, which it mostly does
already but hasn't been stress-tested with concurrent writers.
- **Grafana** — only if the Console's own Prometheus-backed sparklines
turn out to be insufficient once Prometheus ships.
- **"Generalize later" extraction** — the original decision was personal-
first, generalize-later (see Week 1). Once patterns stabilize, extract
a config-driven Oikos core with no `hubris.network`/`hubris`/`strong`
hardcoding, so it's installable on a different homelab.
The Oikos operating model now lives at [`.agents/OIKOS.md`](.agents/OIKOS.md).

View File

@@ -84,4 +84,5 @@ When you change a node:
## See also
- [`CONTRIBUTING.md`](CONTRIBUTING.md) — page templates and tone
- [`.agents/shared/page-templates.md`](.agents/shared/page-templates.md) — page templates and tone
- [`.agents/shared/writing-style.md`](.agents/shared/writing-style.md) — prose style, banned vocabulary

View File

@@ -0,0 +1,164 @@
# 2026-07-06 — Adopt the wiki-hq documentation architecture
## Goal
Reorganize the Homelab-Docs **narrative layer** into the `wiki-hq`
(`/Users/dtoro/Downloads/wiki-hq-main`) documentation model — `sources / wiki / index / log`
plus an `.agents/` separation and a single lint-checkable writing-style standard — **without
breaking** the Oikos machine-readable substrate that reads fixed paths. Keep the existing Mermaid
topology generator (no LikeC4).
## Context
Homelab-Docs is already a mature operational-docs system: `inventory.yaml` as source of truth,
generated `hosts/*.yaml`, an MCP server, a `homelab` CLI, and the Oikos OODA kernel (scheduler,
drift, decide, ledger, approvals). Its weakness is on the *narrative* side — the prose layer grew
organically and lacks the discipline wiki-hq shows:
- No enforceable **writing-style** standard (CONTRIBUTING.md has a one-line "voice" note; CAVEMAN.md governs agent *chat*, not docs).
- **Index/README files are inconsistent** — some pure listings, some prose+tables, some with stale counts.
- **No append-only operations log** for doc maintenance — doc changes are visible only in git.
- **Agent instructions and human content are interleaved** at the repo root (AGENTS.md, OIKOS.md, CAVEMAN.md, CONTRIBUTING.md, GLOSSARY.md alongside `containers/`, `infrastructure/`, …).
Decision (confirmed with operator): **full structural adoption**, **keep Mermaid**, adopt all four
borrows — writing-style guide, section-index/README pattern, append-only per-area logs, and
agent-instruction separation.
## Hard constraint: protect the operational substrate
These paths are read programmatically and **must not move** (see `mcp/server.py`, `bin/homelab`, `oikos/`):
- `inventory.yaml` (root) — MCP (`mcp/server.py:36`), CLI (`bin/homelab:36`), scheduler, drift, relations, gen-topology.
- `hosts/*.yaml` (root, generated) — `HOSTS_DIR` (`mcp/server.py:37`); written by `mcp/build_host_files.py`; read/written by `bin/homelab`.
- `oikos/` — kernel code, `oikos/cards/` (`explain` tool, `mcp/server.py:38`), `oikos/state.json`.
- `secrets/`, `secrets-issuance/`, `ssh/`, `scripts/`, `tools/`, `vps/`, `bin/`, `bootstrap.sh`.
Two MCP tools are path-agnostic and survive any narrative reorg: `search_docs` (ripgreps all
`*.md`) and `get_page(path)` (agent supplies a repo-relative path). The **`## Changelog` convention
must stay** on node pages — `get_changelog` (`mcp/server.py:214`) and the ledger parse it. New
`log.md` files are **additive**, not a replacement for per-page changelogs.
## Current → target structure
Restructure the **narrative layer only**; leave the substrate in place.
```
Homelab-Docs/
README.md # human landing page (kept; refreshed to new nav)
AGENTS.md # kept at root (conventional discovery path)
.agents/ # NEW — agent-facing instruction, separated from content
shared/
writing-style.md # NEW — adapted from wiki-hq (homelab voice)
llm-wiki.md # NEW — the sources/wiki/index/log contract for THIS repo
caveman.md # moved from CAVEMAN.md (agent chat mode)
page-templates.md # moved from CONTRIBUTING.md (page templates)
domains/
knowledge/schema.md # contract for the current-state wiki
operations/schema.md # contract for runbooks / investigations / plans
skills/ # runbooks reshaped as SKILL.md (Phase 4)
lifecycle-provision-node/SKILL.md
service-health-check/SKILL.md
...
OIKOS.md # operating-model doc (moved; agent-facing)
knowledge/ # durable current-state wiki (Feedback/authoritative layer)
index.md # pure listing → section indexes only
log.md # append-only doc-maintenance operations log
GLOSSARY.md # moved from root
sources/
index.md # flat catalog table
references/ # external docs (from infrastructure/references/)
wiki/
hosts/ README.md + hubris.md + strong.md
containers/ README.md (= today's index.md) + <id>-<name>.md
vms/ README.md + <id>-<name>.md
infrastructure/ README.md + grouped: network/ identity/ storage/ ingress/ operations/
operations/ # operator runbook narrative index
index.md commands.md agent-enrollment.md hermes-agent.md
investigations/ # incident evidence (Observe sources)
README.md + log.md + YYYY-MM-DD-*.md + archive/
plans/ # projects / design docs (Act)
README.md + YYYY-MM-DD-*.md + done/
# substrate — UNCHANGED (see Hard constraint)
inventory.yaml hosts/*.yaml oikos/ mcp/ secrets/ secrets-issuance/
ssh/ scripts/ tools/ vps/ bin/ bootstrap.sh
```
**hosts/ split:** generated `hosts/*.yaml` stay at root (substrate); only the two *narrative* pages
`hosts/hubris.md` and `hosts/strong.md` move to `knowledge/wiki/hosts/`. This is the one directory
where machine and narrative content currently mix.
## Consumers to update when narrative paths move
Same phase as the move:
- **`inventory.yaml` `doc_page` fields** — per-host pointer; consumed by `bin/homelab` (`die("no doc_page recorded…")` near `bin/homelab:1288`). Repoint to `knowledge/wiki/...`.
- **`oikos/gen-topology.py` / `oikos/gen_topology_lib.py`** — write `infrastructure/topology.md` and link `oikos/cards/` to doc pages. Move `topology.md` under `knowledge/wiki/infrastructure/` and update the output-path constant (`oikos/gen-topology.py:3`, `:49`, `:195`).
- **`oikos/cards/` templates** and `oikos/drift.py` docstring at `oikos/drift.py:218` (`containers/121-caddy.md`) — cosmetic, update for accuracy.
- **Internal cross-links** — repo-wide relative-link rewrite + link-check pass.
- **Same-session update rule** (`CONTRIBUTING.md:125`) — rewrite its path checklist to the new layout.
`search_docs`, `get_page`, `get_changelog` need **no** code change.
## Writing-style standard (adapted, not copied)
Create `.agents/shared/writing-style.md` from
`/Users/dtoro/Downloads/wiki-hq-main/.agents/shared/writing-style/writing-style.md`:
- **Keep:** banned-vocabulary list (puffers, "leverage/utilize/delve", poetic nouns, promotional adjectives), information-density rule, bold-leading-phrase pattern, tables-over-bullets, `>` blockquotes for caveats, one-sided cross-references, "cross-references are content not catalog".
- **Adjust voice:** wiki-hq mandates strict third-person "no you". Homelab docs are operator runbooks that already use imperative recipes. Allow **imperative/second-person in runbooks, recipes, and skills**; third-person reference voice for node/infrastructure descriptions. State this exception explicitly.
- **Clarify vs CAVEMAN.md:** Caveman governs agent *chat responses*; writing-style governs *committed docs*. No conflict.
- **Make it lint-checkable:** the Oikos lint surface (or a new `.agents/skills/docs-lint`) greps banned words + structural violations.
## Reconciling "append-only per-area logs" with the Oikos ledger
- The **Oikos change ledger** (`oikos/ledger.py`) stays authoritative for *infrastructure changes* (who/what/risk/approval/verification). Do not duplicate it.
- **Per-page `## Changelog`** stays (parsed by `get_changelog`).
- New **per-area `log.md`** (`knowledge/log.md`, `investigations/log.md`, `plans/log.md`) records *doc-maintenance operations only* — restructures, source ingests, lint sweeps — in wiki-hq's single-line format `## [YYYY-MM-DD] <op> | <summary>`.
## Convention: plans always live in `plans/`
The restructure must codify — in `.agents/domains/operations/schema.md`, `AGENTS.md`, and
`.agents/shared/page-templates.md` — that **any plan or design doc for the Homelab is always
written into the repo `plans/` folder** (`plans/YYYY-MM-DD-slug.md`), never into a scratch/agent
plan path, an ad-hoc location, or a chat message. Agents drafting a plan create the file under
`plans/`, list it in `plans/index.md`, and move it to `plans/done/` on completion. This rule is
stated once in the operations schema and cross-referenced from `AGENTS.md` so every agent sees it
at orientation.
## Section-index / README pattern
Every folder gets a `README.md`: 13 sentence prose intro, then a single two-column nav table
`| Document | What it covers |`, nothing else — no stale counts, no duplicated prose. Pure-listing
`index.md` files (e.g. `containers/index.md`) become the folder `README.md`. Root `README.md` stays
the human landing page (may exceed the strict pattern).
## Phased rollout
Each phase independently valuable, independently verifiable, committed separately.
1. **Conventions first (no moves).** Add `.agents/shared/writing-style.md`, `llm-wiki.md`; move CAVEMAN.md → `.agents/shared/caveman.md`, CONTRIBUTING.md → `.agents/shared/page-templates.md` (leave thin root stubs pointing to the new locations). Verify: links resolve; MCP untouched.
2. **Agent-instruction separation.** Move OIKOS.md → `.agents/OIKOS.md`; author `.agents/domains/{knowledge,operations}/schema.md`. Keep AGENTS.md at root, pointing at `.agents/`. Verify: AGENTS.md still the discovery entry; `tools/setup-hermes-soul.sh:51` references resolve or are updated.
3. **Narrative move into `knowledge/wiki/`.** Move `containers/`, `vms/`, `infrastructure/`, host narratives; add `knowledge/index.md`, `knowledge/log.md`, `knowledge/sources/`. Same commit: update `inventory.yaml` `doc_page` fields, `oikos/gen-topology.py` output path, all internal links. Verify: link-check clean; CLI `doc_page` lookups work; `gen-topology.py` writes to the new path; MCP `get_page`/`search_docs` return moved pages.
4. **Runbooks → skills.** Reshape `runbooks/*.md` into `.agents/skills/<name>/SKILL.md` (they already carry `risk_class`, `inputs`, `verification` frontmatter — near-compatible). Update runbook-path references. Verify: frontmatter parses; lifecycle transitions still map to `oikos/ontology.yaml`.
5. **Style + README pass.** Apply the writing-style standard and README pattern across narrative pages; add `log.md` entries recording the restructure. Verify: docs-lint clean.
## Verification (end to end)
- **Link integrity:** markdown link checker (or `rg` over `](` targets) after each move phase; zero broken relative links.
- **MCP resolves docs:** `HOMELAB_CONTEXT_DIR=<repo>``get_page("knowledge/wiki/containers/104-gitea.md")` returns content; `search_docs("gitea")` hits new paths; `get_changelog(<moved page>)` parses.
- **CLI doc_page lookups:** exercise `bin/homelab`'s `doc_page` read against a moved page — no "no doc_page recorded" error.
- **Topology generation:** run `oikos/gen-topology.py` — writes the new `topology.md` path; `oikos/cards/` links point at moved pages.
- **Substrate untouched:** `mcp/build_host_files.py` regenerates `hosts/*.yaml` identically; `git diff` shows no substrate churn.
- **Style lint:** banned-vocabulary grep over `knowledge/` returns no hits; every folder `README.md` matches the intro-plus-nav-table shape.
## Post-migration
Update `plans/index.md` to list this plan; on completion move it to `plans/done/`. Record the
restructure in the new `knowledge/log.md`. Update `README.md` and `AGENTS.md` navigation to the new
layout.
```