Dakota ISO — Skill Router#
Agent entry point for projectbluefin/dakota-iso. Load only the skill(s) that match your task.
Scope: skills here cover work done in this repo. Org-level skills and factory operating model live in
projectbluefin/common/docs/.
Task → Skill#
| I need to... | Load |
|---|---|
| First time / onboarding | |
| Set up a dev environment or clone this repo | docs/skills/onboarding.md |
| Understand issue labels and lifecycle | docs/skills/label-workflow.md |
| Know when to stop and ask a human | docs/skills/human-gates.md |
| Building ISOs | |
| Build ISOs locally, disk space, BTRFS/XFS quirks | docs/build.md |
| Unified nvidia ISO — size, compression, nvidia_imgref | docs/build.md |
Add or modify variants (payload_ref pattern) | docs/variants.md |
| Architecture | |
| Two-container pipeline, boot flow, squashfs, VFS storage | docs/architecture.md |
| GPT layout, El Torito, systemd-boot, dmsquash-live | docs/architecture.md |
| CI/CD | |
build-iso.yml, smoke test, R2 uploads, unified ISO pipeline | docs/ci.md |
| LUKS E2E test (local QEMU, libvirt, CI-equivalent) + installed-disk boot | docs/luks-testing.md |
| R2 / Release | |
| Promoting ISOs to production, rclone, named releases | docs/r2-promotion.md |
| Factory / org | |
| Cross-repo agent rules, branch targets, PR policy | common: docs/factory/agentic-model.md |
| Org structure, parity matrix, open gaps | common: docs/factory/README.md |
| Skill improvement | |
| Writing a skill update alongside a PR | docs/skills/skill-improvement.md |
Skill index#
For the full list of skills in this repo, see docs/skills/INDEX.md.
How to add a skill#
If a task surfaces a non-obvious pattern, workaround, or convention:
- Create (or update) the relevant file in
docs/skills/ - Add a dated section:
### <what you learned> (YYYY-MM-DD) - What failed → why → the fix → code example
- Commit in the same PR as your change — never a follow-up
See docs/skills/skill-improvement.md for the full mandate.
