@keighleykodric/weeve 0.1.1 → 0.1.3

package/README.md

@@ -1,8 +1,8 @@
 # 🧵 Weeve
 
-Cross-functional alignment system. 14+ lanes that audit strategy, product, design, engineering, QA, security, marketing, sales, and UX — then synthesize into one prioritized roadmap.
+Cross-functional contract framework. Each domain writes findings in the same versioned schema shape — Pilot reads them all and surfaces what conflicts, what blocks the ship, and what to do first. Observability for AI-orchestrated execution.
 
-Markdown-in-git. BYO everything. Each lane writes a `<lane>-recommendations.md` file in the same shape. Pilot reads them all and produces a single roadmap. The alignment property: findings from different domains compose, rather than sit in separate files nobody reads.
+Markdown-in-git. BYO everything. The contract shape is the durable asset: a versioned schema that every lane writes to and Pilot synthesizes from. Findings from different domains compose because they share a common structure — not because a platform forces alignment.
 
 ---
 
@@ -10,9 +10,9 @@ Markdown-in-git. BYO everything. Each lane writes a `<lane>-recommendations.md`
 
 1. Install the lanes you need
 2. Run each lane against your product (`/<lane>-intro` to start)
-3. Each lane writes findings to `docs/<lane>/<lane>-recommendations.md`
-4. Run `/pilot-roadmap` — Pilot reads all lane outputs and synthesizes a prioritized roadmap
-5. Run `/pilot-ship` — autonomous worker picks tasks from the roadmap and ships them
+3. Each lane writes findings to `docs/<lane>/<lane>-signals.md`
+4. Run `/pilot-priorities` — Pilot reads all lane outputs and synthesizes a prioritized backlog
+5. Run `/pilot-ship` — autonomous worker picks tasks from the priorities file and ships them
 
 ---
 
@@ -66,9 +66,13 @@ Markdown-in-git. BYO everything. Each lane writes a `<lane>-recommendations.md`
 
 ---
 
+> **Trust model:** Skills execute with your OS permissions — only install from trusted sources.
+
+---
+
 ## You don't need every lane
 
-Install only the lanes that match your team. If marketing doesn't use Maven, that's fine — Maven won't appear in the Pilot roadmap, and nothing breaks.
+Install only the lanes that match your team. If marketing doesn't use Maven, that's fine — Maven won't appear in the Pilot priorities file, and nothing breaks.
 
 Departments without a lane can still contribute. Any team member can feed context into another lane's intake skill — paste a doc, a ticket link, or a screenshot into `/compass-intake` or `/layers-plus-intake` and it gets scoped and routed. You don't need a lane installed to provide input; you just won't get lane-specific findings back.
 
@@ -96,6 +100,16 @@ npx @keighleykodric/weeve add all            # Everything
 
 Every preset includes Pilot — it's the synthesis layer that makes the lanes compose.
 
+### Version pinning
+
+Pin to a tagged release for reproducible installs:
+
+```
+npx skills add keighleykodric/weeve-compass@v0.1.0
+```
+
+See [Releases](https://github.com/keighleykodric/weeve/releases) for available versions.
+
 ---
 
 ## First-run order
@@ -106,22 +120,64 @@ See [WORKFLOW.md](WORKFLOW.md) for the full cadence, within-lane sequences, and
 
 ---
 
-## The alignment property
+## The contract shape
 
-Each lane writes findings independently. Pilot reads them all and surfaces:
+Each lane writes findings independently, in the same schema. Pilot reads them all and surfaces:
 - Cross-lane conflicts (strategy says X, design says Y)
 - Ship gates (Guard or Verify can block `/pilot-ship`)
 - Priority synthesis (what to do first, across all domains)
 
-The claim: running all lanes against the same product produces coherent, composable output — not just a bigger folder of tools.
+The schema is what makes the lanes composable. A finding from Guard and a finding from Compass use the same severity vocabulary, status field, and ID prefix — so Pilot can synthesize them without translation. The contract is the moat, not the lane count.
 
 ---
 
 ## Schema
 
-All lane recommendations files follow a versioned schema: [recommendations-schema.md](docs/shared/recommendations-schema.md) (v0.3).
+All lane signals files follow a versioned schema: [signals-schema.md](spec/schema.md) (v0.4).
+
+The schema is the contract. Severity uses `🔴 High / 🟡 Medium / 🟢 Low`. Lane files never use Now/Next/Watch — that vocabulary belongs to Pilot's priorities output. The manifest hierarchy (`weeve.yaml` → `road.yaml` → `lane.yaml`) governs multi-lane configuration.
+
+---
+
+## Alternative installation (git clone)
+
+If you prefer not to use the npm installer, you can clone lanes directly and verify integrity:
+
+```bash
+# Clone a specific tagged release
+git clone --branch v0.1.0 --depth 1 https://github.com/keighleykodric/weeve-compass.git
+
+# Verify the commit SHA matches the published release
+cd weeve-compass
+git rev-parse HEAD
+# Compare against the SHA listed on https://github.com/keighleykodric/weeve-compass/releases/tag/v0.1.0
+
+# Verify skills-lock.json integrity (if present)
+# skills-lock.json contains SHA-256 hashes of all SKILL.md files
+cat skills-lock.json
+# Manually verify: shasum -a 256 skills/<skill-dir>/SKILL.md
+```
+
+Then symlink or copy the skills directory into `~/.claude/skills/`:
+
+```bash
+cp -r skills/* ~/.claude/skills/
+```
+
+This bypasses the npm registry entirely. You can audit all code before installation and verify that file contents match the published hashes in `skills-lock.json`.
+
+---
+
+## Terminology
+
+Two terms in Weave overlap with common engineering vocabulary:
+
+| Term | In Weave | In engineering contexts |
+|---|---|---|
+| **Operator** | The practitioner running the lanes — a cross-functional lead, designer, or strategist using Claude Code | Infrastructure operator, DevOps engineer, Kubernetes operator |
+| **Schema** | The governance spec that defines how lanes write findings (`spec/schema.md`) | Database schema, OpenAPI spec, JSON Schema |
 
-The schema is the durable asset. Severity uses `🔴 High / 🟡 Medium / 🟢 Low`. Lane files never use Now/Next/Watch — that vocabulary belongs to Pilot's roadmap output.
+If you come from an infrastructure or data engineering background: Weave's "Operator" is closer to "practitioner" or "user," and Weave's "Schema" is closer to "contract format" or "output spec."
 
 ---