agent-loop-ext
Crash-resilient external agent loop with state persistence and CI/CD integration
AUTO-INVOKE for steward feature-domain routing — authoring an expansion (extension/addon/framework), creating or selecting a persona/SOUL identity, or scaffolding a project. Thin discover anchors for the three domains the framework quickrefs don't own.
$ npx -y skills add jmagly/aiwg --skill steward-quickref --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/steward-quickrefContext preview
The summary Claude sees to decide when to auto-load this skill.
AUTO-INVOKE for steward feature-domain routing — authoring an expansion (extension/addon/framework), creating or selecting a persona/SOUL identity, or scaffolding a project. Thin discover anchors for the three domains the framework quickrefs don't own.
namespace: aiwg name: steward-quickref platforms: [all] kernel: true description: AUTO-INVOKE for steward feature-domain routing — authoring an expansion (extension/addon/framework), creating or selecting a persona/SOUL identity, or scaffolding a project. Thin discover anchors for the three domains the framework quickrefs don't own. triggers: - author an expansion - create a persona - select a persona - scaffold a project - steward routing - steward capabilities - steward model routing - model catalog routing - cheap first model policy - which aiwg feature do i use - expansion persona or project
Thin routing anchor for the three cross-cutting **feature domains** that no single framework quickref owns: **expansion authoring**, **persona/SOUL identity**, and **project creation**. This file holds *guidance + anchoring phrases only* — it never inlines skill or agent bodies. The capability index does the heavy lifting: each row gives the `aiwg discover` phrase that ranks the owning capability top-3 (the four discover facets fuse the result — see `src/artifacts/discover-facets.ts`).
> **Always discover, then show.** These phrases are the entry points; fetch the > selected artifact with `aiwg show <type> <name>`. Do not browse the > filesystem (`skill-discovery` rule).
`aiwg-steward.md` is the Tier-1 dispatch core. This quickref is Tier 2. When the steward needs detailed lookup tables, route to Tier 3 with `aiwg discover "aiwg-steward routing reference"` or `aiwg discover "aiwg-steward worked examples"`. If a route is ambiguous, ask one clarifying question; if a target is stale or missing, file a correction issue through the AIWG issue-filing flow with the broken target and observed command output.
"How do I *build* an AIWG expansion?" — extensions, addons, and frameworks.
| You want to… | Discover phrase | Owning capability | |---|---|---| | Author an extension | `aiwg discover "author an expansion"` | `scaffold-extension` | | Create an addon | `aiwg discover "create an addon"` | `scaffold-addon` | | Build a framework | `aiwg discover "scaffold a framework"` | `scaffold-framework` | | Pilot then graduate a project-local bundle | `aiwg discover "project-local bundle"` | `new-bundle`, `promote` | | Package a standalone/community plugin repository | `aiwg discover "standalone plugin repository"` | `package-plugin` |
"How do I *make* or *choose* an identity/voice?" Personas live at `agentic/code/agents/personas/`; SOUL profiles are managed by the `soul-*` skills.
| You want to… | Discover phrase | Owning capability | |---|---|---| | Create / author a persona (SOUL) | `aiwg discover "create a persona"` | `soul-create` | | Select / switch a persona identity | `aiwg discover "select a persona"` | persona agents + `soul-apply` | | See the persona roster | `aiwg discover "persona"` | `agentic/code/agents/personas/*` | | Manage SOUL lifecycle (enable/disable/blend) | `aiwg discover "soul"` | `soul-enable`, `soul-disable`, `soul-blend` |
> Runtime persona *selection UX* (catalog-pick) is an open research area — see > spike #1625; for now, discover the persona then activate via the SOUL skills.
"How do I *start* an AIWG project?"
| You want to… | Discover phrase | Owning capability | |---|---|---| | Scaffold a new project | `aiwg discover "scaffold a project"` | `new-project` | | Create a project-local bundle | `aiwg discover "project-local bundle"` | `new-bundle` |
"Does my provider support X natively, or does AIWG emulate it?" Route through the steward, which reads `capability-matrix.yaml`.
| You want to… | Discover phrase / route | | --- | --- | | Check provider support for a capability | `aiwg discover "provider capability"` → `steward` |
"Which model should this use?" Route through `aiwg steward models` and the `aiwg models` command family. Do not recommend exact provider IDs from memory.
| You want to… | Discover phrase / route | |---|---| | Explain cheap-first routing | `aiwg steward models` | | Inspect effective catalog provenance | `aiwg models sources --json` | | Refresh provider/account catalog | `aiwg models refresh --json` | | Compile/audit artifacts for a provider | `aiwg models audit --provider <provider>` | | Resolve selected model IDs | `aiwg models resolve --provider <provider>` |
When a request maps to one of these domains:
1. **Volunteer the affordance** (Norman signifier): if a user is working near a domain but hasn't found it, surface it — "you can also author expansions, create/select a persona, or scaffold a project; want me to discover one?" 2. **Discover, don't dead-end.** Run the domain's `aiwg discover` phrase. 3. **Re-query on low confidence.** If the first pass is weak, broaden the phrase or try an adjacent domain phrase before concluding "not found" — the `skill-discovery` rule forbids decline-without-search. 4. **Show the selection.** `aiwg show <type> <name>` for the chosen capability.
Reusable project context and specialist workflows for the AI tools you already use. Plan software, coordinate specialist reviews, prepare campaigns, investigate incidents, organize research, curate media, and maintain operational knowledge.
Repo: jmagly/aiwg
Crash-resilient external agent loop with state persistence and CI/CD integration
Detect requests for iterative autonomous agent loops and route to the appropriate loop executor
Automatically execute tests when code-generating agents modify source files, enforcing the execute-before-return pattern
Enable agent loops to learn from similar past tasks and share patterns across loops
Query and manage the executable feedback debug memory
Execute tests on generated code and iterate until passing