prj-unattended-progress-com.../AGENTS.md
codex e231d0fcc6 docs(agents): repoint remote State Hub URL to the in-cluster address
The remote row pointed at 127.0.0.1:18000, a reverse tunnel back to the
workstation. On railiance01 the State Hub runs in the cluster on that same
machine, so the request left the box and came back to reach a local service.

Refs CUST-WP-0067-T07

Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
2026-08-25 00:21:36 +02:00

108 lines
3 KiB
Markdown

# prj-unattended-progress-company — Agent Instructions
## Repo Identity
**Purpose:** Temporary cross-repository project to make Binky Hedgehog GmbH a
profitable, self-organizing company with minimal CEO time and governance.
**Domain:** infotech
**Repo slug:** `prj-unattended-progress-company`
**Workplan prefix:** `UPC-WP-`
**Primary durable partner:** `binky-control` (company control plane / canon)
---
## State Hub Integration
The Custodian State Hub tracks work across domains. Prefer HTTP REST and the
`statehub` / `custodian` CLI. MCP (`dev-hub`) is optional.
| Context | URL |
|---------|-----|
| Local workstation | `http://127.0.0.1:8000` |
| Remote (railiance01, in-cluster) | `http://10.43.68.154:8000` |
Canonical health endpoint: `/state/health` (not `/health`).
### Orient at session start
```bash
# Offline brief when present
cat .custodian-brief.md 2>/dev/null || true
# Project goal (authoritative outcome)
cat GOAL.md
# Active workplans for this repo (after registration)
curl -s "http://127.0.0.1:8000/repos/prj-unattended-progress-company/dispatch" \
| python3 -m json.tool
# Inbox
curl -s "http://127.0.0.1:8000/messages/?to_agent=prj-unattended-progress-company&unread_only=true" \
| python3 -m json.tool
```
### Log progress (required at session close)
```bash
curl -s -X POST http://127.0.0.1:8000/progress/ \
-H "Content-Type: application/json" \
-d '{
"summary": "what was done",
"event_type": "note",
"author": "agent",
"workplan_id": "<uuid-or-omit>",
"task_id": "<uuid-or-omit>"
}'
```
After workplan file changes:
```bash
statehub fix-consistency
```
---
## Session Protocol
**Start:**
1. Read `GOAL.md` — outcome, invariants, success gates, retirement.
2. Read `SCOPE.md` — authority boundary and participating repos.
3. Scan `workplans/` for `ready` / `active` / `blocked` items.
4. Prefer advancing Green/Blue work; package Yellow; batch Orange; never
invent Red-lane authority.
**During work:**
- Implementation lands in participating repos, not here.
- Link child workplans by stable ID; do not duplicate their task lists.
- Record binding cross-repo decisions as work records (decision docs /
hub decisions) with evidence.
**Close:**
1. Update workplan task statuses in files.
2. Hand off residuals as live work records before finishing a workplan.
3. `POST /progress/` with a concrete summary.
4. `statehub fix-consistency` when workplan files changed.
---
## Security
Before requesting credentials, use `warden route find "<need>" --json` when
available. Never record secret values in project files, State Hub, workplans,
evidence, or chat.
<!-- REPO-AGENTS-EXTENSIONS -->
## Project protocol
This is a temporary **project** repository. It has `GOAL.md`, not `INTENT.md`.
Read order at session start:
1. `GOAL.md`
2. `SCOPE.md`
3. Active files under `workplans/`
4. Relevant sections of `binky-control` only when advancing a linked gate
Complete only when every `GOAL.md` retirement condition is met with evidence.