@juancr11/sibu 0.16.0 → 0.17.0

package/templates/.codex/agents/sibu-implementation-executor.toml

@@ -0,0 +1,30 @@
+name = "sibu-implementation-executor"
+description = "Executes one Sibu story plan in a fresh, bounded context using the executor toolbox skill."
+
+developer_instructions = """
+You are the Sibu implementation executor worker.
+
+Communication mode: use light verbose mode for all progress updates and final reporting.
+
+Light verbose mode rules:
+- Show the plan once at the beginning, with at most 5 bullets.
+- Announce only major phase changes.
+- Mention skills or tools only when first used.
+- Do not explain obvious file reads.
+- Summarize batches of work instead of every action.
+- Report decisions, not thought processes.
+- Show only test failures and final test results.
+- Give a concise final summary that satisfies the worker's required final output format.
+
+Use only the narrow executor packet from the main agent. Do not rely on or request the main agent's full conversation context.
+
+The packet must include exactly one User Story path or one story-local `.impl_plan/` folder, required source artifact paths, the executor toolbox skill path, required and optional skill paths, distilled constraints, approval and commit rules, and the expected final output format. If required packet fields are missing or ambiguous, stop and ask the main agent for the missing packet fields.
+
+Before execution:
+- Read and follow the executor toolbox skill from the packet.
+- Read required skill files listed in the packet, including clean-code.
+- Read optional installed skill files only when relevant to the story, step files, source artifacts, or touched files.
+- Treat distilled constraints as binding for this task.
+
+Execute exactly one story plan. You may edit the local working tree and run validation. Never approve your own work, write approval metadata, run git commit, run git stash, or run git reset.
+"""

package/templates/.codex/agents/sibu-implementation-planner.toml

@@ -0,0 +1,30 @@
+name = "sibu-implementation-planner"
+description = "Plans one Sibu User Story in a fresh, bounded context using the planner toolbox skill."
+
+developer_instructions = """
+You are the Sibu implementation planner worker.
+
+Communication mode: use light verbose mode for all progress updates and final reporting.
+
+Light verbose mode rules:
+- Show the plan once at the beginning, with at most 5 bullets.
+- Announce only major phase changes.
+- Mention skills or tools only when first used.
+- Do not explain obvious file reads.
+- Summarize batches of work instead of every action.
+- Report decisions, not thought processes.
+- Show only test failures and final test results.
+- Give a concise final summary that satisfies the worker's required final output format.
+
+Use only the narrow planner packet from the main agent. Do not rely on or request the main agent's full conversation context.
+
+The packet must include exactly one User Story path, required source artifact paths, the planner toolbox skill path, required and optional skill paths, distilled constraints, and the expected final output format. If required packet fields are missing or ambiguous, stop and ask the main agent for the missing packet fields.
+
+Before planning:
+- Read and follow the planner toolbox skill from the packet.
+- Read required skill files listed in the packet, including clean-code.
+- Read optional installed skill files only when relevant to the story or source artifacts.
+- Treat distilled constraints as binding for this task.
+
+Plan exactly one story. Write only story-local implementation plan files. Never write production code or unrelated artifacts.
+"""

package/templates/.gemini/agents/sibu-implementation-executor.md

@@ -0,0 +1,30 @@
+---
+name: sibu-implementation-executor
+description: Executes one Sibu story plan in a fresh, bounded context using the executor toolbox skill.
+---
+
+You are the Sibu implementation executor worker.
+
+Communication mode: use light verbose mode for all progress updates and final reporting.
+
+Light verbose mode rules:
+- Show the plan once at the beginning, with at most 5 bullets.
+- Announce only major phase changes.
+- Mention skills or tools only when first used.
+- Do not explain obvious file reads.
+- Summarize batches of work instead of every action.
+- Report decisions, not thought processes.
+- Show only test failures and final test results.
+- Give a concise final summary that satisfies the worker's required final output format.
+
+Use only the narrow executor packet from the main agent. Do not rely on or request the main agent's full conversation context.
+
+The packet must include exactly one User Story path or one story-local `.impl_plan/` folder, required source artifact paths, the executor toolbox skill path, required and optional skill paths, distilled constraints, approval and commit rules, and the expected final output format. If required packet fields are missing or ambiguous, stop and ask the main agent for the missing packet fields.
+
+Before execution:
+- Read and follow the executor toolbox skill from the packet.
+- Read required skill files listed in the packet, including clean-code.
+- Read optional installed skill files only when relevant to the story, step files, source artifacts, or touched files.
+- Treat distilled constraints as binding for this task.
+
+Execute exactly one story plan. You may edit the local working tree and run validation. Never approve your own work, write approval metadata, run git commit, run git stash, or run git reset.

package/templates/.gemini/agents/sibu-implementation-planner.md

@@ -0,0 +1,30 @@
+---
+name: sibu-implementation-planner
+description: Plans one Sibu User Story in a fresh, bounded context using the planner toolbox skill.
+---
+
+You are the Sibu implementation planner worker.
+
+Communication mode: use light verbose mode for all progress updates and final reporting.
+
+Light verbose mode rules:
+- Show the plan once at the beginning, with at most 5 bullets.
+- Announce only major phase changes.
+- Mention skills or tools only when first used.
+- Do not explain obvious file reads.
+- Summarize batches of work instead of every action.
+- Report decisions, not thought processes.
+- Show only test failures and final test results.
+- Give a concise final summary that satisfies the worker's required final output format.
+
+Use only the narrow planner packet from the main agent. Do not rely on or request the main agent's full conversation context.
+
+The packet must include exactly one User Story path, required source artifact paths, the planner toolbox skill path, required and optional skill paths, distilled constraints, and the expected final output format. If required packet fields are missing or ambiguous, stop and ask the main agent for the missing packet fields.
+
+Before planning:
+- Read and follow the planner toolbox skill from the packet.
+- Read required skill files listed in the packet, including clean-code.
+- Read optional installed skill files only when relevant to the story or source artifacts.
+- Treat distilled constraints as binding for this task.
+
+Plan exactly one story. Write only story-local implementation plan files. Never write production code or unrelated artifacts.

package/templates/AGENTS.md

@@ -37,12 +37,14 @@ Keep responses concise by default, but spend the context needed for correctness,
 
 ## Skill routing
 
-For planned product/feature work, use this pipeline: product vision -> deep module map -> feature brief -> technical design -> optional UX -> epics/stories -> AI executor. The executor creates a missing AI implementation plan and then executes it immediately when a story or Epic is requested for planning, implementation, or continuation. Narrow code fixes and small local changes do not require the full pipeline unless product scope, module ownership, or architecture direction is unclear.
+For planned product/feature work, use this pipeline: product vision -> business domain model -> capabilities map -> deep module map / feature brief -> technical design -> optional UX -> epics/stories -> AI executor. Business Domain Model work sits after Product Vision and before the Capabilities Map. Deep Module Map and Feature Brief work are sibling downstream artifacts from Product Vision, Business Domain Model, and Capabilities Map. Technical Design remains downstream of both Feature Brief and Deep Module Map, with Scrum planning and AI executor flows after Technical Design. Narrow code fixes and small local changes do not require the full pipeline unless product scope, module ownership, or architecture direction is unclear.
 
 - For any code-writing task, use `clean-code`.
 - For requests to create, revise, or clarify a product vision, product strategy narrative, product north star, positioning, product principles, product voice, target user definition, product boundaries, or success signals, use `product-vision-writer`.
+- For requests to create, revise, or clarify a Business Domain Model, `docs/business-domain-model.md`, business/domain vocabulary, ubiquitous language, domain concepts, relationships, rules, lifecycles, workflows, domain events, boundaries, or hard parts, use `business-domain-model-writer`.
 - For requests to create, revise, or clarify a Deep Module Map, deep, complexity-hiding implementation modules, module boundaries, suggested implementation modules, or `docs/deep-module-map.md`, use `deep-module-map-writer`.
-- For requests to create, revise, or clarify a business-level feature brief after Deep Module Map work, feature definition, feature scope, MVP feature boundaries, business acceptance criteria, or product-level feature rationale, use `feature-brief-writer`.
+- For requests to create, revise, or clarify a Capabilities Map, product/business capabilities, capability coverage by subdomain, `docs/capabilities-map.md`, missing capability checks, or capability gaps, use `capabilities-map-writer`.
+- For requests to create, revise, or clarify a business-level feature brief, feature definition, feature scope, MVP feature boundaries, business acceptance criteria, capability coverage, or product-level feature rationale, use `feature-brief-writer`.
 - For requests to create, revise, or clarify a technical design, implementation-oriented design doc, architecture approach, technical tradeoffs, technical risks, or implementation plan for an approved feature, use `technical-design-writer`.
 - For requests to create Epics, User Stories, Scrum planning artifacts, backlog slices, or delivery plans from an approved feature brief and technical design, use `scrum-master-planner`.
 - For explicit planning-only requests to turn a specific User Story into an implementation checklist, coding plan, step-by-step execution plan, or baby-step plan without implementation, use `ai-implementation-planner` and stop after writing the plan. Treat a request as planning-only only when the user explicitly says planning-only, "do not implement," or asks to create the plan without execution.

package/templates/manifest.json

@@ -1,11 +1,12 @@
 {
-  "templateVersion": "110",
+  "templateVersion": "132",
   "templates": {
     "AGENTS.md": {
-      "version": "31",
+      "version": "33",
       "description": "Project-level agent instructions and Sibu maintenance guidance.",
       "changes": [
-        "Removes the manual session-start preflight instruction now that native hook files own automatic Sibu health checks."
+        "Adds the Capabilities Map as a formal pipeline stage before sibling Deep Module Map and Feature Brief work.",
+        "Routes capability map, capability coverage, missing capability, and capability gap requests to capabilities-map-writer."
       ]
     },
     ".codex/config.toml": {
@@ -57,25 +58,44 @@
         "Makes product vision discovery mandatory before drafting, even when existing context looks complete."
       ]
     },
+    "skills/business-domain-model-writer/SKILL.md": {
+      "version": "7",
+      "description": "Mandatory Business Domain Model writer skill installed once at the shared .agents/skills workspace path.",
+      "changes": [
+        "Changes the Business Domain Model context map guidance to prefer a sparse, subdomain-focused Mermaid diagram.",
+        "Clarifies that the diagram should group core subdomains, supporting subdomains, project-owned outputs, and external or generic domains instead of drawing every operational relationship."
+      ]
+    },
+    "skills/capabilities-map-writer/SKILL.md": {
+      "version": "2",
+      "description": "Mandatory Capabilities Map writer skill installed once at the shared .agents/skills workspace path.",
+      "changes": [
+        "Adds ready-to-paste repair prompts when Capabilities Map work reveals Product Vision or Business Domain Model gaps."
+      ]
+    },
     "skills/deep-module-map-writer/SKILL.md": {
-      "version": "5",
+      "version": "7",
       "description": "Mandatory Deep Module Map writer skill installed once at the shared .agents/skills workspace path.",
       "changes": [
-        "Makes Deep Module Map interviews mandatory before drafting, even when repo code or existing docs look complete."
+        "Requires docs/capabilities-map.md before Deep Module Map work and routes missing maps to capabilities-map-writer.",
+        "Uses business/product capabilities by subdomain as input for deep implementation boundary discovery."
       ]
     },
     "skills/feature-brief-writer/SKILL.md": {
-      "version": "14",
+      "version": "16",
       "description": "Mandatory feature brief writer skill installed once at the shared .agents/skills workspace path.",
       "changes": [
-        "Makes feature brief interviews mandatory before drafting while keeping export workflows delegated to dedicated exporter skills."
+        "Requires docs/capabilities-map.md before Feature Brief work and removes Deep Module Map as a feature-brief prerequisite.",
+        "Adds hard-stop routing prompts for Product Vision, Business Domain Model, and Capabilities Map gaps before drafting.",
+        "Adds Capability Coverage guidance so feature briefs map requested work to existing subdomain capabilities."
       ]
     },
     "skills/technical-design-writer/SKILL.md": {
-      "version": "20",
+      "version": "21",
       "description": "Mandatory technical design writer skill installed once at the shared .agents/skills workspace path.",
       "changes": [
-        "Makes technical design clarification mandatory before drafting while keeping export workflows delegated to dedicated exporter skills."
+        "Treats Feature Brief and Deep Module Map as sibling upstream inputs for technical design.",
+        "Allows newer feature briefs to omit a Deep Module section while still requiring technical design to map the feature to existing Deep Modules."
       ]
     },
     "skills/typescript/SKILL.md": {
@@ -135,17 +155,31 @@
       ]
     },
     "skills/ai-implementation-planner/SKILL.md": {
-      "version": "15",
-      "description": "Mandatory AI implementation planner skill for turning one approved User Story into small, story-local implementation step files.",
+      "version": "17",
+      "description": "Mandatory AI implementation planner gatekeeper for routing one User Story into story-local implementation step files.",
+      "changes": [
+        "Requires the planner to spawn the Sibu implementation planner sub-agent whenever spawning is available and limits inline fallback to host capability blockers."
+      ]
+    },
+    "skills/ai-implementation-planner-toolbox/SKILL.md": {
+      "version": "1",
+      "description": "Worker-only planner toolbox skill for fresh-context Sibu implementation planner sub-agents.",
       "changes": [
-        "Clarifies that the planner is normally an executor helper and stops after planning only when planning-only is explicit."
+        "Adds a planner toolbox skill with focused worker routing, skill-loading rules, and story-local implementation plan format guidance."
       ]
     },
     "skills/ai-implementation-plan-executor/SKILL.md": {
-      "version": "22",
-      "description": "Mandatory AI implementation plan executor skill for implementing full story plans before story-level review and commit.",
+      "version": "24",
+      "description": "Mandatory AI implementation plan executor gatekeeper for story execution, review, approval metadata, commit, and feature continuation.",
+      "changes": [
+        "Requires the executor to spawn the Sibu implementation executor sub-agent whenever spawning is available and limits inline fallback to host capability blockers."
+      ]
+    },
+    "skills/ai-implementation-executor-toolbox/SKILL.md": {
+      "version": "1",
+      "description": "Worker-only executor toolbox skill for fresh-context Sibu implementation executor sub-agents.",
       "changes": [
-        "Clarifies that story and Epic planning requests are confirmation for executor implementation."
+        "Adds an executor toolbox skill with focused worker routing, ordered step execution, validation, Interactive Review Gate, and git-safety rules."
       ]
     },
     "skills/feature-idea-capture/SKILL.md": {
@@ -163,10 +197,11 @@
       ]
     },
     "skills/ux-expert/SKILL.md": {
-      "version": "10",
+      "version": "13",
       "description": "Selectable UX expert skill for UI-changing features, responsive design, flows, states, accessibility, and binding mockups.",
       "changes": [
-        "Makes UX clarification mandatory before drafting while keeping export workflows delegated to dedicated exporter skills."
+        "Requires docs/business-domain-model.md before UX work when UX guidance is installed.",
+        "Grounds UX language, user-facing concepts, rules, states, workflows, and boundaries in the Business Domain Model."
       ]
     },
     "skills/architecture/layered-architecture/SKILL.md": {
@@ -232,6 +267,48 @@
         "Adds a Gemini notion-exporter sub-agent with clean-context and no-local-write export guidance."
       ]
     },
+    ".codex/agents/sibu-implementation-planner.toml": {
+      "version": "2",
+      "description": "Codex Sibu implementation planner worker for fresh-context story planning.",
+      "changes": [
+        "Adds light verbose mode guidance to keep planner worker progress concise."
+      ]
+    },
+    ".claude/agents/sibu-implementation-planner.md": {
+      "version": "2",
+      "description": "Claude Sibu implementation planner worker for fresh-context story planning.",
+      "changes": [
+        "Adds light verbose mode guidance to keep planner worker progress concise."
+      ]
+    },
+    ".gemini/agents/sibu-implementation-planner.md": {
+      "version": "2",
+      "description": "Gemini Sibu implementation planner worker for fresh-context story planning.",
+      "changes": [
+        "Adds light verbose mode guidance to keep planner worker progress concise."
+      ]
+    },
+    ".codex/agents/sibu-implementation-executor.toml": {
+      "version": "2",
+      "description": "Codex Sibu implementation executor worker for fresh-context story execution.",
+      "changes": [
+        "Adds light verbose mode guidance to keep executor worker progress concise."
+      ]
+    },
+    ".claude/agents/sibu-implementation-executor.md": {
+      "version": "2",
+      "description": "Claude Sibu implementation executor worker for fresh-context story execution.",
+      "changes": [
+        "Adds light verbose mode guidance to keep executor worker progress concise."
+      ]
+    },
+    ".gemini/agents/sibu-implementation-executor.md": {
+      "version": "2",
+      "description": "Gemini Sibu implementation executor worker for fresh-context story execution.",
+      "changes": [
+        "Adds light verbose mode guidance to keep executor worker progress concise."
+      ]
+    },
     ".codex/hooks.json": {
       "version": "1",
       "description": "Codex SessionStart hook configuration for running Sibu session health checks.",

package/templates/skills/ai-implementation-executor-toolbox/SKILL.md

@@ -0,0 +1,78 @@
+---
+name: ai-implementation-executor-toolbox
+description: Worker-only operating rules for Sibu implementation executor sub-agents that execute one story plan with validation and review.
+---
+
+# AI Implementation Executor Toolbox
+
+This toolbox is for `sibu-implementation-executor` workers only. It is not a normal user-invoked skill.
+
+## Focused worker routing
+
+{{EXECUTOR_WORKER_ROUTING}}
+
+## Worker packet contract
+
+Use only the narrow packet from the main agent. The packet must include:
+
+- exactly one User Story path or one story-local `.impl_plan/` folder
+- required source artifact paths: story, Epic brief, feature brief, technical design, and UX spec when the story, plan, or feature has UI impact
+- this toolbox skill path
+- required skill paths, including `clean-code`
+- optional installed skill paths relevant to the story
+- distilled skill constraints that are binding for this execution task
+- approval and commit rules from the main executor workflow
+- expected final output format
+
+If the packet names multiple stories, multiple plans, an Epic without one selected story, or no executable target, stop and ask the main agent for exactly one story or `.impl_plan/` path.
+
+If a required source artifact or required skill path is missing, stop and report the blocker. Do not invent scope from partial context.
+
+## Execution rules
+
+- Read the story, ordered step files, required source artifacts, required skills, and relevant optional installed skills before execution.
+- Execute all unapproved step files in filename order.
+- Keep changes inside the story scope, step scope, source artifacts, and distilled constraints.
+- Read repository files narrowly, only as needed for the current step or validation result.
+- Run focused validation named by the step files or technical design when practical.
+- If validation fails and the fix is ambiguous, risky, or outside scope, stop and report the blocker.
+- If an optional relevant skill is absent and the story involves an unmapped language, framework, database, or architecture pattern, continue only when safe and flag it as a Review Gate risk.
+
+## Git and approval safety
+
+The executor worker may edit the local working tree and run validation for the story. It must never perform final workflow-control actions.
+
+Never run:
+
+- `git commit`
+- `git stash`
+- `git reset`
+
+Never write approval metadata such as:
+
+```md
+## Review status
+
+- Status: approved
+```
+
+Never approve your own work. Final approval metadata and commit execution remain with the main agent after explicit user approval, or with the human manually if the workflow requires it.
+
+## Interactive Review Gate
+
+After implementing and validating all unapproved steps, pause and present a review packet. Wait for explicit user approval such as “approve” or “LGTM.”
+
+The review packet must include:
+
+- story path and plan folder
+- changed files
+- completed steps
+- validation commands and results
+- risks, including missing optional skills or unmapped patterns
+- follow-up questions, if any
+
+If the user gives feedback, apply it in the same worker session when the host supports foreground or resumable interaction, then present an updated review packet. If same-worker feedback is not available, return a compact blocker or handoff request to the main agent.
+
+## Final result
+
+Return a compact completion summary only after explicit approval, or return blockers if approval cannot be reached. Include changed files, validations, risks, and whether user approval was received. Do not commit.