palladius/attila

آخر نشاط مصدر مسجل: ٢ يوليو ٢٠٢٦ في ٠٨:٥٩

آخر تحديث لفهرس SkillsMP: ١٢ نوفمبر ٢٠٢٦ في ١٢:٥٢

skills مجمعة: ١

نجوم GitHub: ٠
تفرعات GitHub: ٠

التثبيت بمساعدة مساعد ذكي

انسخ هذا الطلب إلى المساعد الذكي الذي تستخدمه.

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/palladius/attila.

ملف README للمستودع

مصدر ملف README للمستودع: README.md

قراءة الملف الأصلي

عرض ملف README كاملًا

# A.TT.I.L.A. (Flagello di Dio) 🗡️ "A dockerized, safe to use version of the SRE Extension, good for discovery and investigations". <p align="center"> <img src="assets/images/attila_sre_squad.jpg" width="400" alt="Attila's Spappari SRE Squad Logo"> </p> > *"A come atroce, T come terremoto, T come tragedia, I come iradiddio, L come laco de sangue, A come... come Attila!"* — Diego Abatantuono Project **A.TT.I.L.A.** is a stateful Site Reliability Engineering (SRE) investigation and cloud discovery tool on Google Cloud Platform (GCP). It enables Gemini-managed agents to retain context, memory, and state across runs by using a persistent storage layer. --- ## 🗡️ Barbarian SREs at Work Here is a visual representation of the **Spappari SRE Horde** invading a motherboard to perform system diagnostics, troubleshooting, and mitigations: <p align="center"> <img src="assets/images/barbarian_horde_sre.jpg" width="600" alt="Spappari SRE Motherboard Invasion"> </p> --- ## 🏗️ Architecture Overview The system consists of the setup/orchestration CLI (`attila`) and the barbarian agent harness container (`spapparo`). ![Architecture Infographic](assets/images/attila_architecture_infographic.jpg) ### Core Workflow: 1. **CLI setup**: The host machine invokes the Python CLI to set up workspace environments. 2. **Container Launch**: Launches the Docker container `spapparo` which mounts a `/memory` directory to persist rules and reports. 3. **IAM Impersonation**: The container impersonates a restricted GCP Service Account (`safe-sre-investigator`) with a read-only `Viewer` role to ensure safety during audits. 4. **Vertex AI Discovery**: The container uses `@google/gemini-cli` in YOLO mode (`-y`) to call Vertex AI, run discovery command sequences, and output findings back to `/memory`. --- ## 🚀 Getting Started ### 1. Initialize the project workspace Initialize the directory structure and create the configuration files for your GCP project: ```bash just init <PROJECT_ID> ``` *Alternatively, call the CLI directly:* ```bash python3 cli/attila.py init --project-id <PROJECT_ID> ``` This generates the directory layout in `memory/<PROJECT_ID>/`: - `discovery/` - Logs of past discovery runs. - `rules/` - Prompt rules that govern agent behavior. - `investigations/` - Logs of SRE incident response activities. - `.env` - Autodetects your current `gcloud` identity. ### 2. Set Up Google Cloud Infrastructure Ensure APIs are enabled and provision the restricted Service Account and GCS Buckets. **Via Terraform (Recommended):** ```bash just setup-infra-tf ``` **Via Bash Script (Bypasses local restrictions):** ```bash just setup-infra-bash ``` ### 3. Build & Run the Container Build the barbarian container: ```bash just docker-build ``` Execute a discovery run: ```bash just run-discovery "Perform a discovery of GCS Buckets in project $PROJECT_ID" ``` --- ## 🧪 Evaluations & Testing We validate agent behavior using LLM-as-a-judge evaluations matching expected resource outputs. - **Fast Unit Tests**: ```bash just test ``` - **Run Agentic Evals**: ```bash python3 run_evals.py ``` This loads definitions from `tests/evals.yaml`, triggers a discovery run, and grades findings using Gemini against expected resources. --- ## 🗺️ Roadmap & Milestones ``` ┌────────────────────────────────┐ │ v0.1 PoC (Local Storage) │ │ - Local bind-mounts │ │ - --harness geminicli │ │ - Bash/TF infra provisioning │ └───────────────┬────────────────┘ │ ▼ ┌────────────────────────────────┐ │ v0.2 Target (Cloud State) │ │ - --storage gcs │ │ - --harness adk │ │ - Incident 'Tragedies' tracking│ └───────────────┬────────────────┘ │ ▼ ┌────────────────────────────────┐ │ v0.3 Target (Mitigation) │ │ - Alert/PubSub triggers │ │ - Guided incident mitigation │ │ - Slack/CLI human-in-the-loop│ └────────────────────────────────┘ ``` See [MILESTONES.md](docs/MILESTONES.md) and [SPECS.md](docs/SPECS.md) for deeper details.

جارٍ تحميل ملف README للمستودع…

Skills في هذا المستودع

محللو أمن المعلومات

١ فئات مهنية · 100% مصنفة

skill

المهنة

الوصف

آخر تحديث

discovery-gcp-benjamin

محللو أمن المعلومات

Perform structured GCP resource discovery and security auditing using sequential gcloud queries, generating a JSON catalog and a security Wiki.

لغة النص الأصلي: الإنجليزية

2026-07-02

عرض ١ من أصل ١ skills مجمعة.