feat(v1.5.0): prompt-pack maturity — all packs to standard + CI enforcement

docs/PROMPT_INDEX.md

@@ -1,7 +1,7 @@
 # Atlas One — Prompt Index
 
-Version: 1.3.0
-Generated: 2026-06-07
+Version: 1.4.0
+Generated: 2026-07-02
 
 This index is auto-generated by `scripts/generate-prompt-index.sh`.
 Do not edit manually — run the script to regenerate.

docs/prompts/packs/architecture-review.md

@@ -11,9 +11,9 @@ This pack is for active design review and critique, not retrieval.
 
 ## Recommended Modes
 
-- Architect
-- Review
-- Analyze
+* Architect
+* Review
+* Analyze
 
 ## Starter Prompt — Full Architecture Review
 
@@ -98,3 +98,22 @@ Architect
 Output:
 Architecture summary, risks, open questions, one structural recommendation.
 ```
+
+## When to Use
+
+* Before a design hardens — when you can still change structure cheaply
+* When reviewing someone else's architecture, or your own with fresh eyes
+* When a decision feels risky but the risk is not yet named
+
+## Expected Output
+
+* A 2–3 sentence architecture summary
+* Risks ordered by severity, with the reasoning behind each
+* Unstated assumptions made explicit
+* One structural change that improves resilience, plus open questions
+
+## Constraints
+
+* Critique, do not just validate — surface what is missing or wrong
+* Ground findings in the design provided; mark inferences as inferences
+* Review only — implementation and risk gating belong to mq-agent / mq-mcp

docs/prompts/packs/interview-prep.md

@@ -8,9 +8,9 @@ and system design interviews.
 
 ## Recommended Modes
 
-- Plan
-- Coach
-- Teach
+* Plan
+* Coach
+* Teach
 
 ## Starter Prompt — Interview Strategy
 
@@ -98,3 +98,22 @@ Plan
 Output:
 What they're testing, what I must communicate, prep steps in order.
 ```
+
+## When to Use
+
+* When preparing for a specific interview with limited time
+* When you need structured answers, not just a topic list
+* When practicing behavioral, technical or system-design rounds
+
+## Expected Output
+
+* What each stage is testing for
+* The few things you must communicate clearly
+* Ordered preparation steps and what to avoid
+* For practice rounds: a structured answer and one follow-up question
+
+## Constraints
+
+* Keep spoken answers realistic in length (under ~2 minutes)
+* Base answers on the candidate's real experience — no fabricated stories
+* Coaching only; the pack prepares you, it does not sit the interview

docs/prompts/packs/learning-coach.md

@@ -5,11 +5,17 @@ Tags: pack, learning, coaching, teaching, practice
 
 Use this pack to learn a subject through structured explanation and practice.
 
+## When to Use
+
+* When picking up a new subject and you want a practical learning sequence
+* When you understand a topic loosely but can't yet explain or apply it
+* When you want practice tasks and a way to check your own understanding
+
 ## Recommended Modes
 
-- Teach
-- Explain
-- Coach
+* Teach
+* Explain
+* Coach
 
 ## Starter Prompt
 
@@ -25,3 +31,31 @@ Include:
 - one practice task
 - how I can tell if I understood it
 ```
+
+## Expected Output
+
+* A short prerequisite ("understand first") step
+* A plain-language explanation, not jargon
+* One concrete, worked example
+* One practice task the learner can actually do
+* A self-check to confirm understanding
+
+## Constraints
+
+* Explanations stay at the learner's stated level — no unexplained jargon
+* Do not fabricate facts; flag uncertainty instead of guessing
+* Coaching only — no execution, no external accounts or tools required
+
+## Example Usage
+
+```text
+Goal:
+Learn how TLS certificate chains work well enough to debug a broken one.
+
+Mode:
+Teach
+
+Output:
+What to understand first, a plain explanation, a worked example, a practice
+task, and a self-check.
+```

docs/prompts/packs/macos-scripts-terminal.md

@@ -8,16 +8,16 @@ Covers mqlaunch command surface, automation scripts, and workflow structure.
 
 ## Recommended Modes
 
-- Plan
-- Debug
-- Architect
+* Plan
+* Debug
+* Architect
 
 ## When to Use
 
-- When adding a new command to mqlaunch
-- When a terminal workflow breaks or behaves unexpectedly
-- When designing a new automation or login flow
-- When deciding how a new feature fits the macos-scripts command surface
+* When adding a new command to mqlaunch
+* When a terminal workflow breaks or behaves unexpectedly
+* When designing a new automation or login flow
+* When deciding how a new feature fits the macos-scripts command surface
 
 ## Starter Prompt
 
@@ -81,3 +81,30 @@ Notes:
 - See .claude/settings.local.example.json for recommended placeholders and instructions to create your own `.claude/settings.local.json`.
 - Use `MQ_ROOT` or replace `<MQ_ROOT>` placeholders with your repository path.
 ```
+
+## Expected Output
+
+* A clear statement of what the workflow does and when it triggers
+* The existing commands/scripts it touches, and its inputs and outputs
+* Failure modes with graceful handling, and a definition of done
+* For command-surface work: the command, flags, no-arg behavior and help text
+
+## Constraints
+
+* Match existing mqlaunch naming and command-surface conventions
+* One command does one clear thing; fail gracefully with a clear message
+* No secrets or machine-specific private paths; use `MQ_ROOT` placeholders
+
+## Example Usage
+
+```text
+Goal:
+Add an `mqlaunch flow` command that lists and launches saved workflows.
+
+Mode:
+Plan
+
+Output:
+What it does, scripts it touches, inputs/outputs, failure handling, and the
+command-surface shape (command, flags, no-arg behavior, help text).
+```

docs/prompts/packs/mq-ecosystem-boundaries.md

@@ -9,9 +9,9 @@ another's begins.
 
 ## Recommended Modes
 
-- Analyze
-- Architect
-- Decide
+* Analyze
+* Architect
+* Decide
 
 ## Ecosystem Boundary Map
 
@@ -79,3 +79,35 @@ Compare the candidate tools on:
 
 Recommend one owner and explain the boundary clearly.
 ```
+
+## When to Use
+
+* Before designing a workflow that spans more than one mq tool
+* When it is unclear which tool should own a responsibility
+* When a change risks blurring the line between two tools
+
+## Expected Output
+
+* For each tool: what it owns, what it does not, and its handoffs
+* What breaks if a given tool is removed or fails
+* The one critical boundary the design must respect
+
+## Constraints
+
+* Atlas One provides reasoning structure, not execution
+* Do not re-implement mq-mcp review logic or bypass mq-agent's safety layer
+* Respect each tool's stated authority; flag ambiguous ownership
+
+## Example Usage
+
+```text
+Goal:
+Design where repo-signal ends and mq-mcp begins for a release-readiness flow.
+
+Mode:
+Analyze
+
+Output:
+Per-tool responsibilities and handoffs, failure impact, and the critical
+boundary to respect.
+```

docs/prompts/packs/mq-mcp-architecture-memory.md

@@ -9,16 +9,16 @@ structured memory records into actionable architectural insight.
 
 ## Recommended Modes
 
-- Architect
-- Analyze
-- Decide
+* Architect
+* Analyze
+* Decide
 
 ## When to Use
 
-- When reading ADRs from mq-mcp architecture_memory/
-- When a new change may conflict with an existing architecture decision
-- When designing a new component and needing to check established boundaries
-- When onboarding to the mq ecosystem and needing to understand system philosophy
+* When reading ADRs from mq-mcp architecture_memory/
+* When a new change may conflict with an existing architecture decision
+* When designing a new component and needing to check established boundaries
+* When onboarding to the mq ecosystem and needing to understand system philosophy
 
 ## Starter Prompt
 
@@ -74,3 +74,29 @@ Draft an ADR that includes:
 - Consequences
 - Constraints this creates for future work
 ```
+
+## Expected Output
+
+* What constraint or principle the ADR establishes, and why
+* The boundary it defines and what breaks if it is ignored
+* A verdict for the current change: compatible / conflicts / needs clarification
+
+## Constraints
+
+* Reason from the retrieved ADR text; do not invent decisions
+* This pack interprets memory — it does not write to architecture_memory/ directly
+* New ADR drafts are proposals for human review, not accepted records
+
+## Example Usage
+
+```text
+Goal:
+Check a new provider design against an existing "mq-mcp owns validation" ADR.
+
+Mode:
+Architect
+
+Output:
+The constraint the ADR sets, the boundary it defines, and a compatible /
+conflicts / needs-clarification verdict for the change.
+```

docs/prompts/packs/mq-mcp-safety-review.md

@@ -9,15 +9,15 @@ thinking that makes mq-mcp review more effective.
 
 ## Recommended Modes
 
-- Review
-- Analyze
-- Decide
+* Review
+* Analyze
+* Decide
 
 ## When to Use
 
-- Before running a mq-mcp security or risk review on a repo
-- When deciding whether a change is safe to approve
-- When assessing a new MCP tool or server for safety boundaries
+* Before running a mq-mcp security or risk review on a repo
+* When deciding whether a change is safe to approve
+* When assessing a new MCP tool or server for safety boundaries
 
 ## Starter Prompt
 
@@ -74,7 +74,27 @@ State:
 ## Safety Constraints
 
 Do not use this pack to:
-- Bypass mq-mcp approval gates
-- Approve changes that request hidden chain-of-thought
-- Approve tool execution that circumvents user confirmation
-- Substitute for a real mq-mcp security review where one is required
+* Bypass mq-mcp approval gates
+* Approve changes that request hidden chain-of-thought
+* Approve tool execution that circumvents user confirmation
+* Substitute for a real mq-mcp security review where one is required
+
+## Expected Output
+
+* Findings ordered by severity (critical / warn / info), each with evidence
+* Recommended fixes or constraints per finding
+* An explicit go / no-go recommendation with conditions
+
+## Example Usage
+
+```text
+Goal:
+Prepare a safety brief before running mq-mcp review on a new MCP server.
+
+Mode:
+Review
+
+Output:
+Severity-ordered findings with evidence, recommended constraints, and a
+go / no-go recommendation.
+```