personal-projects/leo-claude-mktplace
Template
0
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(agents): add permissionMode, disallowedTools, skills frontmatter to all 25 agents #409

Merged
lmiranda merged 1 commits from feat/agent-frontmatter-hardening-v3 into development 2026-02-03 16:10:25 +00:00
Conversation 0 Commits 1 Files Changed 28 +201 -87
lmiranda commented 2026-02-03 16:07:41 +00:00
Owner
Copy Link

Summary

Comprehensive agent-level configuration using Claude Code's supported frontmatter fields (permissionMode, disallowedTools, skills).

Changes:

  • permissionMode added to all 25 agents:

    • bypassPermissions (1): Executor — full autonomy with code-sentinel + Code Reviewer safety nets
    • acceptEdits (7): Orchestrator, Data Ingestion, Theme Setup, Refactor Advisor, Doc Analyzer, Git Assistant, Maintainer
    • default (7): Planner, Code Reviewer, Data Advisor, Layout Builder, Full Validation, Clarity Coach, CMDB Assistant
    • plan (10): All pr-review agents (5), Data Analysis, Design Reviewer, Component Check, Agent Check, Security Reviewer (code-sentinel)

  • disallowedTools added to 12 agents:

    • All plan-mode agents (10) + Code Reviewer + Clarity Coach receive disallowedTools: Write, Edit, MultiEdit
    • Enforces read-only contracts at platform level (defense-in-depth)

  • Model promotions:

    • Planner: sonnet → opus (architectural reasoning)
    • Code Reviewer: sonnet → opus (quality gate thoroughness)

  • skills frontmatter on 3 agents:

    • Executor: 7 safety-critical skills auto-injected
    • Code Reviewer: 4 review skills auto-injected
    • Maintainer: 2 config skills auto-injected
    • Body text ## Skills to Load removed for these agents to avoid duplication

  • Documentation updated:

    • CLAUDE.md and docs/CONFIGURATION.md with complete agent configuration matrix
    • New subsections: permissionMode Guide, disallowedTools Guide, skills Frontmatter Guide

Test plan

  • Verify validate-marketplace.sh passes
  • Test Executor autonomy with /sprint-start (should proceed without permission prompts)
  • Verify Code Reviewer read-only with /review (should not write files)
  • Test Config Maintainer fluency with /config-audit-settings
  • Verify PR-review isolation with /pr-review (all sub-agents in read-only mode)
  • Assess Planner depth with /sprint-plan (opus should provide deeper analysis)

🤖 Generated with Claude Code

## Summary Comprehensive agent-level configuration using Claude Code's supported frontmatter fields (`permissionMode`, `disallowedTools`, `skills`). **Changes:** - **permissionMode** added to all 25 agents: - `bypassPermissions` (1): Executor — full autonomy with code-sentinel + Code Reviewer safety nets - `acceptEdits` (7): Orchestrator, Data Ingestion, Theme Setup, Refactor Advisor, Doc Analyzer, Git Assistant, Maintainer - `default` (7): Planner, Code Reviewer, Data Advisor, Layout Builder, Full Validation, Clarity Coach, CMDB Assistant - `plan` (10): All pr-review agents (5), Data Analysis, Design Reviewer, Component Check, Agent Check, Security Reviewer (code-sentinel) - **disallowedTools** added to 12 agents: - All `plan`-mode agents (10) + Code Reviewer + Clarity Coach receive `disallowedTools: Write, Edit, MultiEdit` - Enforces read-only contracts at platform level (defense-in-depth) - **Model promotions:** - Planner: `sonnet` → `opus` (architectural reasoning) - Code Reviewer: `sonnet` → `opus` (quality gate thoroughness) - **skills frontmatter** on 3 agents: - Executor: 7 safety-critical skills auto-injected - Code Reviewer: 4 review skills auto-injected - Maintainer: 2 config skills auto-injected - Body text `## Skills to Load` removed for these agents to avoid duplication - **Documentation** updated: - `CLAUDE.md` and `docs/CONFIGURATION.md` with complete agent configuration matrix - New subsections: permissionMode Guide, disallowedTools Guide, skills Frontmatter Guide ## Test plan - [ ] Verify `validate-marketplace.sh` passes - [ ] Test Executor autonomy with `/sprint-start` (should proceed without permission prompts) - [ ] Verify Code Reviewer read-only with `/review` (should not write files) - [ ] Test Config Maintainer fluency with `/config-audit-settings` - [ ] Verify PR-review isolation with `/pr-review` (all sub-agents in read-only mode) - [ ] Assess Planner depth with `/sprint-plan` (opus should provide deeper analysis) 🤖 Generated with [Claude Code](https://claude.ai/code)
lmiranda added the Type/Feature label 2026-02-03 16:07:41 +00:00
lmiranda added 1 commit 2026-02-03 16:07:41 +00:00
feat(agents): add permissionMode, disallowedTools, skills frontmatter to all 25 agents 8ea8a3de59 
- permissionMode: 1 bypassPermissions, 7 acceptEdits, 7 default, 10 plan
- disallowedTools: 12 agents blocked from Write/Edit/MultiEdit
- model: promote Planner + Code Reviewer to opus
- skills: auto-inject on Executor (7), Code Reviewer (4), Maintainer (2)
- docs: CLAUDE.md + CONFIGURATION.md updated with full agent matrix

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
lmiranda force-pushed feat/agent-frontmatter-hardening-v3 from 8ea8a3de59 to 19ba80191f 2026-02-03 16:08:59 +00:00 Compare
lmiranda merged commit a005610a37 into development 2026-02-03 16:10:25 +00:00
lmiranda deleted branch feat/agent-frontmatter-hardening-v3 2026-02-03 16:10:25 +00:00
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
Clear labels
Component/API
API endpoints, contracts, and integration
 Component/Auth
Authentication and authorization
 Component/Backend
Backend service code and business logic
 Component/Database
Database schemas, migrations, queries
 Component/Deploy
Deployment, infrastructure, DevOps
 Component/Docs
Documentation and guides
 Component/Frontend
User interface and client-side code
 Component/Infra
Infrastructure and system configuration
 Component/Testing
Test infrastructure and frameworks
 Tech/Docker
Docker containers and compose
 Tech/FastAPI
FastAPI backend framework
 Tech/JavaScript
JavaScript/Node.js code
 Tech/PostgreSQL
PostgreSQL database
 Tech/Python
Python language and libraries
 Tech/Redis
Redis cache and pub/sub
 Tech/Vue
Vue.js frontend framework
Complexity/Complex
High uncertainty or many dependencies
 Complexity/Medium
Some unknowns or dependencies
 Complexity/Simple
Straightforward, well-understood
 Effort/L
3-5 days
 Effort/M
1-2 days
 Effort/S
2-4 hours
 Effort/XL
> 1 week
 Effort/XS
< 2 hours
 Priority/Critical
Immediate attention required
 Priority/High
Important, needs attention soon
 Priority/Low
Nice to have, can wait
 Priority/Medium
Normal priority
 Type/Bug
Something isn't working
 Type/Chore
Maintenance tasks
 Type/Documentation
Documentation improvements
 Type/Feature
New feature or request
 Type/Refactor
Code refactoring without changing behavior
 Type/Test
Testing related
No Label Type/Feature
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
lmiranda
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: personal-projects/leo-claude-mktplace#409
Delete Branch "feat/agent-frontmatter-hardening-v3"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?