Skip to content

Commit 44e6bd4

Browse files
author
ainish-coder bot
committed
feat(security): Add comprehensive AGENTS.md security protocols and recommendations
- Introduced a new agents.proposed.md file outlining security-first, zero trust, and standardized coding approaches. - Added detailed security principles, data protection measures, and post-quantum cryptography standards. - Created agents_improvement.md to document security protocol improvement recommendations based on recent NIST and OWASP publications. - Updated the PQC table to reflect finalized standards and included new recommendations for key encapsulation and digital signatures. - Enhanced zero trust architecture guidelines and included OWASP LLM & Agentic AI security updates. - Established a migration timeline and priority matrix for implementing recommended changes.
1 parent 3025401 commit 44e6bd4

17 files changed

+3489
-61
lines changed

.gitignore

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -16,6 +16,7 @@
1616
.trae/
1717

1818
# PlatformIO ignore patterns
19+
.pio*
1920
.pio/
2021
*.pioenvs/
2122
pio.ini

.osa-yolo-mode/OSA_CLI_AGENTS.md

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,49 @@
1+
# OSA: Orchestrated System of Agents
2+
3+
## Agent CLI Commands
4+
5+
| Agent | Command | Use Case | Priority |
6+
|-------|---------|----------|----------|
7+
| **Gemini** | `gemini --yolo "prompt"` | Orchestration, planning, merging | 1 |
8+
| **Qwen** | `qwen --yolo "prompt"` | Fast code generation | 2 |
9+
| **OpenCode** | `opencode run "prompt"` | Schema validation, security | 3 |
10+
| **Crush** | `crush run "prompt"` | Security audit, code review | 4 |
11+
| **Claude** | `claude -p "prompt" --dangerously-skip-permissions` | Architecture, QA | 5 |
12+
13+
## Core Rules
14+
15+
1. **Non-Interactive Only**: All commands must run autonomously—no TUI prompts
16+
2. **Parallel Execution**: Use `&` and `wait` for concurrent agent tasks
17+
3. **State Sync**: Agents share state via `.toon` files (MEMORY.toon, TODO.toon)
18+
19+
## Orchestration Patterns
20+
21+
### Parallel Build
22+
```bash
23+
opencode run "Create schema" &
24+
qwen --yolo "Write API routes" &
25+
crush run "Analyze components" &
26+
wait
27+
```
28+
29+
### Debug Chain
30+
1. **Crush**: Analyze codebase → `ANALYSIS.md`
31+
2. **Qwen**: Implement fix
32+
3. **OpenCode**: Verify
33+
34+
### Review Chain
35+
1. **Qwen**: Generate code (speed)
36+
2. **OpenCode**: Audit for security
37+
3. **Gemini**: Merge
38+
39+
## State File Format
40+
```toon
41+
task_id: feat-01
42+
swarm[2]{agent,task,status}:
43+
qwen,api,done
44+
opencode,schema,failed
45+
```
46+
47+
## Safety
48+
- Use `git worktree` for parallel changes
49+
- Abort swarm if >50% agent failure rate

0 commit comments

Comments
 (0)