agent-loop-ext
Crash-resilient external agent loop with state persistence and CI/CD integration
Validate a Flow graph manifest, explain semantic diagnostics, and emit one provider-neutral normalized composition contract.
$ npx -y skills add jmagly/aiwg --skill composition-validate --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/composition-validateContext preview
The summary Claude sees to decide when to auto-load this skill.
Validate a Flow graph manifest, explain semantic diagnostics, and emit one provider-neutral normalized composition contract.
namespace: aiwg name: composition-validate platforms: [all] description: Validate a Flow graph manifest, explain semantic diagnostics, and emit one provider-neutral normalized composition contract. triggers: - validate a Flow graph - validate composition manifest - check graph joins and ceilings - normalize agent tool graph
Use this skill when an operator needs to author or validate **flow.aiwg.io/v1alpha1 FlowGraph**.
1. Confirm the manifest uses **apiVersion: flow.aiwg.io/v1alpha1** and **kind: FlowGraph**. 2. Run **aiwg composition validate MANIFEST --format json**. 3. When an AIWG index export is available, add **--catalog INDEX.json** so stable candidate IDs are proven against that captured catalog. 4. Repair every error diagnostic. Do not weaken permissions, ceilings, approval boundaries, retry modes, or cycle guards merely to pass validation. 5. Pass the returned normalized envelope to an adapter or runtime. Do not add provider-specific fields to the source graph.
chain-of-thought.
See **schemas/flow-graph.schema.json**, **README.md**, and **docs/schema-evolution.md** in the addon for the complete contract.
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