fraim-framework 2.0.66 → 2.0.67

package/dist/registry/ai-manager-rules/user-survey-phases/phase4-survey-build-x.md

@@ -0,0 +1,21 @@
+# Phase 4: Building for X (Twitter)
+
+## Objective
+Generate short-form, high-engagement content for X, including poll copy or threaded questions.
+
+## Key Actions
+1. **Brevity**: Ensure copy fits within character limits (280 for free, or longer for Premium).
+2. **Engagement Hook**: Create a concise "Question of the Day" format.
+3. **Poll Duration**: Recommend optimal poll duration (e.g., 24 hours or 7 days).
+4. **Visuals**: Suggest or generate an accompanying image if a poll isn't used.
+
+## Validation Criteria
+- [ ] Copy is within character limits.
+- [ ] Poll options are clear and 25 characters or less.
+- [ ] Content is "retweetable" (viral potential).
+
+## Evidence Required
+- **X Survey Doc**: `docs/surveys/51-x-survey.md`
+
+## Next Steps
+Proceed to Phase 5: Building for Facebook.

package/dist/registry/ai-manager-rules/user-survey-phases/phase5-survey-build-facebook.md

@@ -0,0 +1,19 @@
+# Phase 5: Building for Facebook
+
+## Objective
+Generate content for Facebook Groups or Profiles, focusing on community engagement and personal networking.
+
+## Key Actions
+1. **Group Identification**: List target Facebook Groups.
+2. **Contextual Hook**: Frame the survey as a "seeking advice" post for the community.
+3. **Admin Check**: Remind the user to check with Group Admins if required.
+
+## Validation Criteria
+- [ ] Tone is helpful and collaborative.
+- [ ] Post includes a clear prompt for comments.
+
+## Evidence Required
+- **Facebook Survey Doc**: `docs/surveys/51-facebook-survey.md`
+
+## Next Steps
+Proceed to Phase 6: Custom Platform Build.

package/dist/registry/ai-manager-rules/user-survey-phases/phase6-survey-build-custom.md

@@ -0,0 +1,15 @@
+# Phase 6: Custom Platform Build
+
+## Objective
+Assess the need for additional platforms (Discord, Slack, Instagram) and generate custom copy accordingly.
+
+## Key Actions
+1. **Identify Gaps**: Review if the core 4 platforms reach the target audience effectively.
+2. **Custom Format**: Adapt copy for Slack/Discord (markdown-heavy) or Instagram (visual-heavy).
+3. **Execution Plan**: Define how these supplementary surveys should be dispatched.
+
+## Validation Criteria
+- [ ] User confirms no more platforms are needed OR custom copy is generated.
+
+## Next Steps
+Proceed to Phase 7: Dispatch & Automated Execution.

package/dist/registry/ai-manager-rules/user-survey-phases/phase7-survey-dispatch.md

@@ -0,0 +1,45 @@
+# Phase 7: Dispatch & Automated Execution
+
+## Objective
+Execute the survey distribution across the prepared platforms using browser automation (Playwright).
+
+## Agent Instructions (Playwright)
+
+**CRITICAL**: Use Playwright MCP tools to interact with platforms. Do NOT rely on hard-coded selectors. Use your intelligence to find buttons and fields by role or label.
+
+### General Dispatch Process
+1. **Navigate**: Go to the platform URL.
+2. **Login Verification**: Check if the user is logged in.
+   - If NOT logged in: **Pause and ask the user to log in manually** in the browser.
+3. **Drafting**: Locate the "Start a post" or "Create post" button.
+4. **Filling**: Enter the copy generated in Phases 2-5.
+5. **Poll Creation**: Click the "Add Poll" button if applicable and fill in options.
+6. **Execution**: **Pause and ask for user confirmation** before clicking "Post".
+
+## Platform-Specific Guidance
+
+### LinkedIn
+- Look for `button[name="Start a post"]`.
+- Click "Create a poll" icon.
+- Enter questions and options.
+
+### Reddit
+- Navigate to the specific subreddit URL (e.g., `reddit.com/r/SaaS/submit`).
+- Select "Poll" or "Post" tab.
+- Enter title and body.
+
+### X (Twitter)
+- Locate the "Post" button or the large "What is happening?!" textbox.
+- Click the "Poll" icon.
+
+## Validation Criteria
+- [ ] Post successfully drafted on the target platform.
+- [ ] Screenshot of the "Ready to Post" state shared with the user.
+- [ ] User has explicitly clicked "Confirm" or given permission to click "Post".
+
+## Evidence Required
+- **Screenshot**: `docs/evidence/screenshots/[platform]-ready-to-post-{timestamp}.png`
+- **Execution Log**: Mentioned in the chat.
+
+## Next Steps
+The User Survey Dispatch workflow is complete once all platforms are posted.

package/dist/registry/delivery/address-feedback.md

@@ -0,0 +1,83 @@
+# Delivery Phase: Address Feedback
+
+## INTENT
+Capture feedback, address all items, then re-validate work through the workflow's quality gates before resubmitting.
+
+## OUTCOME
+Either:
+- **Approved**: No unresolved feedback, proceed to retrospective
+- **Has Feedback**: Feedback captured, addressed, and work re-validated through workflow
+
+## WORKFLOW
+
+### Step 1: Wait for Feedback
+
+{{delivery.await_feedback}}
+
+### Step 2: Check Feedback
+
+{{delivery.check_feedback}}
+
+### Step 3: Evaluate Results
+
+**If work is approved and no unresolved feedback exists:**
+- Call the Mentor with
+
+```javascript
+seekMentoring({
+  workflowType: "{workflow_type}",
+  issueNumber: "{issue_number}",
+  currentPhase: "address-feedback",
+  status: "complete",
+  evidence: {
+    approved: true,
+    feedbackRounds: "{total_rounds}"
+  }
+})
+```
+
+**If there is feedback or changes requested:**
+- Proceed to Step 4
+
+### Step 4: Write Feedback File
+
+**CRITICAL**: Write feedback file to `docs/evidence/{issueNumber}-{workflowType}-feedback.md`
+
+Create or append to the feedback file:
+
+```markdown
+# Feedback for Issue #{issueNumber} - {WorkflowType} Workflow
+
+## Round {X} Feedback
+*Received: {timestamp}*
+
+{{delivery.feedback_item_format}}
+```
+
+**Round Detection Logic**:
+1. If feedback file doesn't exist → Start with "Round 1"
+2. If feedback file exists → Parse to find highest "Round X", use "Round X+1"
+3. Always append new round, never overwrite existing rounds
+
+### Step 5: Address Each Feedback Item
+
+- Work through each item systematically
+- Make the required code/document changes
+- Mark each feedback item as ADDRESSED in the feedback file
+{{delivery.address_commit_note}}
+
+### Step 6: REPORT BACK
+
+```javascript
+seekMentoring({
+  workflowType: "{workflow_type}",
+  issueNumber: "{issue_number}",
+  currentPhase: "address-feedback",
+  status: "failure",
+  findings: {
+    feedbackFile: "docs/evidence/{issue_number}-{workflow_type}-feedback.md",
+    roundNumber: "{X}",
+    itemsAddressed: "{count}"
+  }
+})
+```

package/dist/registry/delivery/retrospective.md

@@ -0,0 +1,78 @@
+# Delivery Phase: Retrospective
+
+## INTENT
+Conduct a comprehensive retrospective of the completed work while full context is available, capturing learnings and insights before cleanup and resolution.
+
+## WORKFLOW
+
+### Step 1: Retrospective Preparation
+- Gather work artifacts (commits, documents, test results, etc.)
+- Review original issue requirements and acceptance criteria
+- **Collect all feedback**: Review all comments, suggestions, and feedback received during review
+- Analyze feedback patterns and themes
+- Identify key decision points and challenges encountered
+
+### Step 2: Create Retrospective Document
+- Create (or use existing) retrospective file: `docs/retrospectives/issue-{issue-number}-{kebab-title}-postmortem.md`
+- Use retrospective template: `get_fraim_file({ path: "templates/retrospective/RETROSPECTIVE-TEMPLATE.md" })`
+- Document while work context is fresh and complete
+
+### Step 3: Work Analysis
+- **What Went Well**: Successful approaches, tools, and techniques used
+- **What Went Poorly**: Challenges, obstacles, and inefficiencies encountered
+- **Root Cause Analysis**: Why problems occurred, not just what happened
+- **Decision Analysis**: Review key decisions made during the work
+- **Feedback Analysis**: What feedback revealed about the work quality
+- **Process Effectiveness**: How well the workflow supported the work
+
+### Step 4: Learning Extraction
+- **Key Learnings**: Specific insights that can help future work of this type
+- **Process Improvements**: Workflow or rule changes that would help
+- **Tool/Technique Discoveries**: New approaches or tools that proved valuable
+- **Anti-Patterns Identified**: Approaches to avoid in future work
+- **Feedback Patterns**: Common themes in feedback to address
+- **Quality Insights**: What contributed to or detracted from work quality
+
+### Step 5: Prevention Measures
+- **Specific Actions**: Concrete steps to prevent similar issues in future work
+- **Rule Updates**: Suggestions for updating FRAIM rules or workflows
+- **Knowledge Sharing**: Insights that should be shared with other agents
+- **Process Enhancements**: Workflow improvements based on experience
+- **Feedback Integration**: How to better incorporate feedback in future work
+- **Quality Improvements**: Changes to improve work quality
+
+### Step 6: Validation and Completion
+- Ensure retrospective meets quality criteria
+- Validate all template sections are complete
+- Confirm learnings are actionable and specific
+- Verify feedback has been analyzed and learned from
+- Check that prevention measures are concrete and implementable
+
+## VALIDATION CRITERIA
+- ✅ Retrospective document created using template
+- ✅ Root cause analysis completed (not just symptoms)
+- ✅ Prevention measures documented and actionable
+- ✅ Key learnings extracted and documented
+- ✅ Process improvements identified
+- ✅ Feedback analyzed and lessons captured
+- ✅ All template sections completed with substantial content
+- ✅ Retrospective quality meets standards
+- ✅ Learnings are specific to the workflow type and transferable
+
+## REPORT BACK
+
+When retrospective is complete:
+
+```javascript
+seekMentoring({
+  workflowType: "{workflow_type}",
+  issueNumber: "{issue_number}",
+  currentPhase: "retrospective",
+  status: "complete",
+  evidence: {
+    retrospectivePath: "docs/retrospectives/issue-{issue_number}-{title}-postmortem.md",
+    learningsDocumented: true,
+    preventionMeasuresIdentified: true
+  }
+})
+```

package/dist/registry/delivery/submit.md

@@ -0,0 +1,79 @@
+# Delivery Phase: Submit for Review
+
+## INTENT
+Submit completed work for review, documenting evidence and presenting work through the appropriate channel.
+
+## OUTCOME
+Work is submitted for review with:
+- Evidence documented
+- Work presented for review
+- Ready for feedback
+
+## WORKFLOW
+
+### Step 1: Create Evidence Document
+
+**File location:**
+```
+docs/evidence/{issue_number}-{workflow_type}-evidence.md
+```
+
+**Evidence document should include:**
+
+#### Summary
+- Issue number and title
+- Workflow type (spec, design, implement, test)
+- Brief description of work completed
+
+#### Work Completed
+- Key files changed
+- Approach taken
+- Testing completed (if applicable)
+
+#### Feedback History (if applicable)
+**CRITICAL**: If feedback file exists (`docs/evidence/{issue_number}-{workflow_type}-feedback.md`), include its complete contents in the evidence document under this section.
+
+#### Validation
+- How work was validated
+- Validation results
+- Screenshots/output (if applicable)
+
+#### Quality Checks
+- All deliverables complete
+- Documentation clear and professional
+- Work ready for review
+
+#### Phase Completion
+- All phases completed for this workflow
+- Evidence from each phase
+- Any iterations or challenges
+
+### Step 2: Sync Work
+
+{{delivery.sync_work}}
+
+### Step 3: Submit for Review
+
+{{delivery.submit_for_review}}
+
+### Step 4: Final Verification
+
+{{delivery.final_checklist}}
+
+## REPORT BACK
+
+When all steps are complete, report back:
+
+```javascript
+seekMentoring({
+  workflowType: "{workflow_type}",
+  issueNumber: "{issue_number}",
+  currentPhase: "submit",
+  status: "complete",
+  evidence: {
+    evidenceDocument: "docs/evidence/{issue_number}-{workflow_type}-evidence.md",
+    workSynced: true,
+    submittedForReview: true
+  }
+})
+```

package/dist/registry/providers/ado.json

@@ -0,0 +1,19 @@
+{
+  "get_issue": "mcp_ado_get_work_item({ organization: '{{repository.organization}}', project: '{{repository.project}}', workItemId: {{issue_number}} })",
+  "update_issue_status": "mcp_ado_update_work_item({ organization: '{{repository.organization}}', project: '{{repository.project}}', workItemId: {{issue_number}}, state: 'Active' })",
+  "create_pr": "mcp_ado_create_pull_request({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', title: '{{title}}', sourceBranch: '{{branch}}', targetBranch: 'main', description: '{{body}}' })",
+  "get_pr": "mcp_ado_get_pull_request({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}} })",
+  "get_pr_comments": "mcp_ado_get_pr_comments({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}} })",
+  "get_pr_review_comments": "mcp_ado_get_pr_review_comments({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}} })",
+  "get_pr_reviews": "mcp_ado_get_pr_reviews({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}} })",
+  "add_pr_comment": "mcp_ado_add_pr_comment({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}}, content: '{{comment}}' })",
+  "merge_pr": "mcp_ado_merge_pull_request({ organization: '{{repository.organization}}', project: '{{repository.project}}', repository: '{{repository.name}}', pullRequestId: {{pr_number}} })",
+  "create_branch": "git checkout -b feature/{{issue_number}}-{{slug}}",
+  "switch_branch": "git checkout {{branch}}",
+  "commit_changes": "git add . && git commit -m '{{message}}'",
+  "push_branch": "git push origin {{branch}}",
+  "list_issues": "mcp_ado_list_work_items({ organization: '{{repository.organization}}', project: '{{repository.project}}', state: 'Active' })",
+  "create_issue": "mcp_ado_create_work_item({ organization: '{{repository.organization}}', project: '{{repository.project}}', workItemType: 'Bug', title: '{{title}}', description: '{{body}}' })",
+  "add_issue_comment": "mcp_ado_add_work_item_comment({ organization: '{{repository.organization}}', project: '{{repository.project}}', workItemId: {{issue_number}}, text: '{{comment}}' })",
+  "assign_issue": "mcp_ado_update_work_item({ organization: '{{repository.organization}}', project: '{{repository.project}}', workItemId: {{issue_number}}, assignedTo: '{{assignee}}' })"
+}

package/dist/registry/providers/delivery-conversation.json

@@ -0,0 +1,9 @@
+{
+  "delivery.sync_work": "**DO NOT commit changes.** The user commits manually in Conversation mode.",
+  "delivery.submit_for_review": "**Present work to user:**\n- Show the user a summary of all changes made\n- Show diff of what changed\n- Suggest a commit message if helpful\n- Wait for user to review",
+  "delivery.final_checklist": "**Checklist:**\n- ✅ Evidence document created and complete\n- ✅ Work presented to user with summary and diff\n- ✅ All deliverables complete",
+  "delivery.await_feedback": "Wait for the user to provide feedback on the work. Do not proceed until user has reviewed.",
+  "delivery.check_feedback": "Evaluate the user's response:\n- Did they approve the work?\n- Did they request changes or provide feedback?",
+  "delivery.feedback_item_format": "### Item {N} - UNADDRESSED\n- **Feedback**: {user_feedback}\n- **Status**: UNADDRESSED",
+  "delivery.address_commit_note": "\n- **DO NOT commit changes** - the user commits manually"
+}

package/dist/registry/providers/delivery-pr.json

@@ -0,0 +1,9 @@
+{
+  "delivery.sync_work": "**Commit all changes:**\n- Stage all modified files\n- Create commit with descriptive message\n- Push to feature branch\n\n**Commit message format:**\n```\n{workflow_type}({issue_number}): complete {phase_name} phase\n\n- Evidence document created\n- All deliverables complete\n- Ready for review\n```",
+  "delivery.submit_for_review": "**Verify PR exists:**\n- PR exists for the feature branch\n- PR title reflects the issue and workflow\n- PR description includes context and links to deliverables\n\n**If PR doesn't exist:** Create PR and link to issue.\n\n**Add PR comment** with link to evidence:\n```\n{Workflow_type} workflow complete.\nEvidence: docs/evidence/{issue_number}-{workflow_type}-evidence.md\nReady for human review.\n```\n\n**Update issue labels:**\n- Remove: `status:wip`\n- Add: `status:needs-review`\n- Keep: Current phase label\n\nUse {{update_issue_status}} to update labels.",
+  "delivery.final_checklist": "**Checklist:**\n- ✅ Evidence document created and complete\n- ✅ All work committed and pushed to feature branch\n- ✅ PR exists and is ready\n- ✅ PR comment added with evidence link\n- ✅ Issue labels updated (status:needs-review)\n- ✅ Git status clean",
+  "delivery.await_feedback": "**Wait for human instruction.** Do not automatically poll or check PR status.\n\nWhen human says \"check the PR\" or \"proceed\", move to Step 2.",
+  "delivery.check_feedback": "Use {{get_pr}} and related actions to check the PR comprehensively:\n\n1. **Get PR Status**: {{get_pr}} - Check if PR is approved\n2. **Get All Comment Types**:\n   - **PR-level comments**: {{get_pr_comments}} - Get general comments\n   - **Review comments**: {{get_pr_review_comments}} - Get inline/detailed comments\n   - **Review status**: {{get_pr_reviews}} - Get approval/changes requested status",
+  "delivery.feedback_item_format": "### Comment {N} - UNADDRESSED\n- **Author**: {author}\n- **Type**: {pr_comment|review_comment|review}\n- **File**: {file_path} (if applicable)\n- **Line**: {line_number} (if applicable)\n- **Comment**: {comment_body}\n- **Status**: UNADDRESSED",
+  "delivery.address_commit_note": ""
+}

package/dist/registry/providers/github.json

@@ -0,0 +1,19 @@
+{
+  "get_issue": "mcp_github_issue_read({ owner: '{{repository.owner}}', repo: '{{repository.name}}', issue_number: {{issue_number}}, method: 'get' })",
+  "update_issue_status": "mcp_github_issue_write({ owner: '{{repository.owner}}', repo: '{{repository.name}}', issue_number: {{issue_number}}, method: 'update', labels: ['status:in-progress'] })",
+  "create_pr": "mcp_github_create_pull_request({ owner: '{{repository.owner}}', repo: '{{repository.name}}', title: '{{title}}', head: '{{branch}}', base: 'main', body: '{{body}}' })",
+  "get_pr": "mcp_github_pull_request_read({ owner: '{{repository.owner}}', repo: '{{repository.name}}', pullNumber: {{pr_number}}, method: 'get' })",
+  "get_pr_comments": "mcp_github_pull_request_read({ owner: '{{repository.owner}}', repo: '{{repository.name}}', pullNumber: {{pr_number}}, method: 'get_comments' })",
+  "get_pr_review_comments": "mcp_github_pull_request_read({ owner: '{{repository.owner}}', repo: '{{repository.name}}', pullNumber: {{pr_number}}, method: 'get_review_comments' })",
+  "get_pr_reviews": "mcp_github_pull_request_read({ owner: '{{repository.owner}}', repo: '{{repository.name}}', pullNumber: {{pr_number}}, method: 'get_reviews' })",
+  "add_pr_comment": "mcp_github_add_issue_comment({ owner: '{{repository.owner}}', repo: '{{repository.name}}', issue_number: {{pr_number}}, body: '{{comment}}' })",
+  "merge_pr": "mcp_github_merge_pull_request({ owner: '{{repository.owner}}', repo: '{{repository.name}}', pullNumber: {{pr_number}} })",
+  "create_branch": "git checkout -b feature/{{issue_number}}-{{slug}}",
+  "switch_branch": "git checkout {{branch}}",
+  "commit_changes": "git add . && git commit -m '{{message}}'",
+  "push_branch": "git push origin {{branch}}",
+  "list_issues": "mcp_github_list_issues({ owner: '{{repository.owner}}', repo: '{{repository.name}}', state: 'OPEN' })",
+  "create_issue": "mcp_github_issue_write({ owner: '{{repository.owner}}', repo: '{{repository.name}}', method: 'create', title: '{{title}}', body: '{{body}}' })",
+  "add_issue_comment": "mcp_github_add_issue_comment({ owner: '{{repository.owner}}', repo: '{{repository.name}}', issue_number: {{issue_number}}, body: '{{comment}}' })",
+  "assign_issue": "mcp_github_issue_write({ owner: '{{repository.owner}}', repo: '{{repository.name}}', issue_number: {{issue_number}}, method: 'update', assignees: ['{{assignee}}'] })"
+}

package/dist/registry/rules/agent-success-criteria.md

@@ -0,0 +1,101 @@
+# Agent Success Criteria & Incentives
+
+## Intent
+To clearly define how AI Agents are evaluated in the FRAIM ecosystem. Agents must optimize for these 5 criteria to be considered successful.
+
+## The 6 Success Criteria
+
+| Criterion | Weight | Definition |
+| :--- | :--- | :--- |
+| **1. Integrity** | **100 pts** | Truthfulness, admitting failures, ethical reporting. |
+| **2. Correctness** | **50 pts** | Adherence to architecture, passing tests, valid code, learning from others. |
+| **3. Contributing to Others' Success** | **40 pts** | Tracking feedback, writing quality retrospectives, helping future agents. |
+| **4. Completeness** | **30 pts** | Thorough execution of workflows, full coverage of requirements. |
+| **5. Independence** | **20 pts** | Smart decision making, learning from history, minimizing noise. |
+| **6. Speed** | **10 pts** | Efficiency and velocity of execution. |
+
+**CRITICAL: Context Preservation** - Part of Independence (#5)
+When working with FRAIM workflows, you MUST maintain workflow context across conversation compaction. See detailed guidance below.
+
+---
+
+## Detailed Guidelines
+
+### 1. Integrity (100 Points) - NON-NEGOTIABLE
+*The Foundation. If you fail here, you fail everywhere.*
+- **Truthfulness**: Never claim a test passed if you didn't run it. Never claim a task is done if it's not.
+- **Admission**: If you cannot do something, say so. If you made a mistake, admit it immediately.
+- **Reporting**: Your status updates must strictly reflect reality.
+- **Violation**: Lying about test results or "hallucinating" success is an immediate failure of the entire task.
+
+### 2. Correctness (50 Points)
+*Doing it right and learning from others.*
+- **Architecture**: Follow the layers defined in `docs/architecture/architecture.md`. No shortcuts.
+- **Testing**: Code requires tests. Tests must pass.
+- **Types**: No `any` types unless absolutely necessary. Solve TypeScript errors, don't suppress them.
+- **Learning from Others**: Check prioritized learnings (via `get_fraim_file({ path: "prioritized-learnings.md" })`) before starting work to avoid known mistakes.
+- **Historical Patterns**: Review similar past issues in `retrospectives/` to apply proven approaches.
+
+### 3. Contributing to Success (40 Points)
+*Helping future agents succeed.*
+- **Feedback Tracking**: Immediately capture all user feedback in evidence documents as it happens.
+- **Quality Retrospectives**: Write detailed, actionable retrospectives with clear learnings and patterns.
+- **Pattern Documentation**: Identify and document patterns that future agents can use to avoid mistakes.
+- **Evidence Quality**: Provide comprehensive evidence that helps others understand what worked and what didn't.
+
+### 4. Completeness (30 Points)
+*Finishing the job.*
+- **Coverage**: Did you handle the edge cases? Did you implement *everything* in the issue description?
+- **Workflow**: Did you follow every step of the workflow? (e.g., did you delete the feature branch after merging?)
+- **Artifacts**: Did you update the docs? Did you update `task.md`?
+
+### 5. Independence (20 Points)
+*Smart decisions, low noise.*
+- **Self-Reliance**: Check the rules and workflows (via `get_fraim_file`) and past issues (`retrospectives/`) *before* asking the user.
+- **Smart Context**: Don't ask "Where is file X?". Use your tools to find it.
+- **Decision Making**: When blocked, propose a solution (Option A vs Option B) rather than just saying "I am blocked."
+- **Context Preservation**: CRITICAL - Maintain FRAIM workflow context across conversation compaction (see below).
+
+#### Context Preservation (CRITICAL for Independence)
+
+When working with FRAIM workflows via `seekMentoring`, you MUST maintain context:
+
+**Track these values throughout the workflow:**
+- `workflowType` - The workflow you're following (e.g., "implement", "spec")
+- `issueNumber` - The issue/task ID (e.g., "#123")
+- `currentPhase` - Where you are in the workflow (e.g., "implement-core")
+
+**After conversation compaction:**
+1. Look for "Starting Workflow: [name]" or "Current Phase: [name]" in your last `seekMentoring` response
+2. Extract the workflow name and phase from these structured phrases
+3. Call `seekMentoring` again with those values and `status="starting"` to resume
+
+**The information is always there:**
+- Workflow name appears in "Starting Workflow: X" or "Workflow: X"
+- Current phase appears in "Current Phase: X" or "Moving to the next phase: X"
+- You provided the issueNumber - keep using the same one
+
+**Example:**
+If your last response said "Current Phase: implement-core", resume with:
+```
+seekMentoring({
+  workflowType: "implement",
+  issueNumber: "#123",
+  currentPhase: "implement-core",
+  status: "starting"
+})
+```
+
+**Why this matters:**
+- Users get frustrated when you forget you're working with FRAIM
+- Losing workflow context breaks the entire process
+- This directly impacts your Independence score - don't ask users to remind you
+
+### 6. Speed (10 Points)
+*Velocity.*
+- **Efficiency**: Use the most direct tool for the job.
+- **Batching**: Validation steps should be batched where possible (run all tests, not just one).
+- **Conciseness**: Keep your reasoning short and actionable.
+
+## Evaluation
+These criteria will be used to evaluate your performance in Retrospectives and future automated gating checks.