@farmslot/skills 0.1.0

package/skills/recipe-harness/SKILL.md

@@ -0,0 +1,31 @@
+---
+name: recipe-harness
+description: Help run or install a local recipe runtime without assuming the full Farmslot OS
+compatibility: Claude, Codex, Cursor, and Markdown skill runners with shell access
+metadata:
+  package: '@farmslot/skills'
+  related_package: '@farmslot/recipe-harness'
+allowed-tools: Read Bash(rg:*) Bash(node:*) Bash(npm:*) Bash(yarn:*) Bash(pnpm:*)
+---
+
+# Recipe Harness
+
+Use the local recipe runtime when one exists, or recommend the smallest install path when it does not.
+
+## Workflow
+
+1. Find the project-owned recipe command first.
+2. If no command exists, check for `@farmslot/recipe-harness` or an app-specific harness package.
+3. Run dry-run or schema validation before live execution when available.
+4. Record artifact paths and validation output.
+5. If no runner exists, stop with a concrete install recommendation.
+
+## Hard Rules
+
+- Do not invent runner commands.
+- Do not require Farmslot Gateway, pool files, slots, Command Center, or Companion for first use.
+- Do not bypass project fixtures or user-flow validation by directly mutating app state.
+
+## Output
+
+Return the command run, exit status, artifacts written, and any next install step.

package/skills/recipe-quality/SKILL.md

@@ -0,0 +1,45 @@
+---
+name: recipe-quality
+description: Review validation recipes and evidence quality without project-specific assumptions
+compatibility: Claude, Codex, Cursor, and Markdown skill runners with read access to recipe artifacts
+metadata:
+  package: '@farmslot/skills'
+  outputs: recipe-quality findings or artifacts/recipe-quality.json when the runner supports artifacts
+allowed-tools: Read Bash(rg:*) Bash(node:*) Bash(git:*)
+---
+
+# Recipe Quality
+
+Review whether a recipe proves the claims it says it proves.
+
+## Use When
+
+- A PR, ticket, or investigation includes a proposed validation recipe
+- Evidence exists but the proof strength is unclear
+- A reviewer needs to know whether the recipe fakes state, skips important assertions, or relies on screenshots when state proof is stronger
+
+## Review Checklist
+
+1. Identify the source claims and acceptance criteria.
+2. Map each recipe node to the claim it proves.
+3. Check that setup, actions, assertions, and teardown are explicit.
+4. Verify that evidence is produced by the real user or system path, not injected state.
+5. Flag unresolved claims instead of stretching the recipe to cover them.
+6. Recommend the smallest recipe delta that would make weak proof reviewable.
+
+## Hard Rules
+
+- Do not accept hidden UI or store mutation as proof of a user flow.
+- Do not mark visual claims as proven without visual evidence or an explicit non-visual contract.
+- Do not mark state claims as proven by screenshots alone when structured state proof is available.
+- Do not require Farmslot Gateway, Command Center, pool files, or slots for a first-pass quality review.
+
+## Output
+
+Return:
+
+- verdict: `pass`, `warn`, or `fail`
+- strongest covered claims
+- weak or missing claims
+- evidence gaps
+- concrete recipe delta

package/templates/recipe-cook/scenarios.json

@@ -0,0 +1,24 @@
+{
+  "version": 1,
+  "scenarios": [
+    {
+      "id": "review-explore-markets-live",
+      "lane": "review",
+      "label": "Explore markets live review",
+      "task_artifact_dir": "projects/example-browser-farm/tasks/review/41802-0416-1517/artifacts"
+    },
+    {
+      "id": "fix-zero-balance-cta",
+      "lane": "fix",
+      "label": "Zero-balance CTA bug fix",
+      "task_artifact_dir": "projects/example-browser-farm/tasks/fix/41692-0413-2142/artifacts"
+    },
+    {
+      "id": "investigation-probe-output",
+      "lane": "investigation",
+      "label": "Probe-output investigation lane",
+      "task_artifact_dir": "projects/example-browser-farm/tasks/review/41705-0414-1243/artifacts",
+      "probe_output_dir": "projects/example-browser-farm/tasks/review/41705-0414-1243/artifacts/probe-output"
+    }
+  ]
+}