argus-sre 0.2.0__tar.gz

argus_sre-0.2.0/.dockerignore

@@ -0,0 +1,8 @@
+.git
+__pycache__
+*.pyc
+*.pyo
+.pytest_cache
+.ruff_cache
+dist
+*.egg-info

argus_sre-0.2.0/.github/workflows/ci.yml

@@ -0,0 +1,66 @@
+name: CI
+
+on:
+  pull_request:
+    branches: [main]
+  push:
+    branches: [main]
+
+jobs:
+  check-commits:
+    name: Conventional Commits
+    runs-on: ubuntu-latest
+    if: github.event_name == 'pull_request'
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - run: pip install commitizen
+      - run: cz check --rev-range origin/${{ github.base_ref }}..HEAD
+
+  lint:
+    name: Lint
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - run: pip install ruff
+      - run: ruff check argus/
+
+  smoke-test:
+    name: Import Smoke Test
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - run: pip install -e .
+      - run: python -c "from argus.main import app; from argus.commands import agent, diagnose, incident; print('OK')"
+
+  docker:
+    name: Build & Push Image
+    runs-on: ubuntu-latest
+    if: github.event_name == 'push' && github.ref == 'refs/heads/main'
+    needs: [lint, smoke-test]
+    permissions:
+      contents: read
+      packages: write
+    steps:
+      - uses: actions/checkout@v4
+      - uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - uses: docker/build-push-action@v5
+        with:
+          context: .
+          push: true
+          tags: ghcr.io/t12-pybash/argus:latest
+          labels: org.opencontainers.image.revision=${{ github.sha }}

argus_sre-0.2.0/.github/workflows/publish.yml

@@ -0,0 +1,21 @@
+name: Publish
+
+on:
+  push:
+    tags:
+      - "v*.*.*"
+
+jobs:
+  publish:
+    name: Publish to PyPI
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - run: pip install hatch
+      - run: hatch build
+      - uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          password: ${{ secrets.PYPI_API_TOKEN }}

argus_sre-0.2.0/.github/workflows/release.yml

@@ -0,0 +1,25 @@
+name: Release
+
+on:
+  push:
+    tags:
+      - "v*.*.*"
+
+jobs:
+  release:
+    name: Build & Publish to PyPI
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      id-token: write
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - name: Install hatch
+        run: pip install hatch
+      - name: Build
+        run: hatch build
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

argus_sre-0.2.0/.gitignore

@@ -0,0 +1,6 @@
+__pycache__/
+*.pyc
+*.pyo
+.venv/
+dist/
+*.egg-info/

argus_sre-0.2.0/CHANGELOG.md

@@ -0,0 +1,44 @@
+## v0.2.0 (2026-06-27)
+
+### Feat
+
+- **diagnose**: RAG over Obsidian vault — search Qdrant for runbooks before LLM call
+- **diagnose**: use deepseek-r1:7b for root cause analysis, strip <think> blocks from output
+- **argus**: add 'argus model' command family (status, rollout, rollback)
+- **status**: add KServe InferenceService health to argus status
+- expand agent remediation action space
+- rename package to argus, add agent command and CI/CD pipeline
+- direct Qdrant upsert on save — bypass vault sync pipeline
+- add sre diagnose command and enrich ask with pod exit reasons
+- add problem pods, deployment readiness, disk usage, and Flux sync context to sre ask
+- inject running pod context into sre ask
+- replace rsync CronJob with git-based vault publisher
+- add CPU and memory context to sre ask with node name mapping
+- add Dockerfile and SRE_CONFIG_FILE support
+- add incremental re-indexing to ingest_vault
+- add sre incident command with vault note creation and cluster snapshot
+- add sre runbook command with semantic vault search and optional LLM synthesis
+- initial sre CLI with status, alert, ask commands
+
+### Fix
+
+- **diagnose**: exit early when pod has no evidence of problems
+- remove unused subprocess import from agent
+- update Dockerfile and CI for argus rename
+- rename all sre references to argus
+- strip bold markers from Slack remediation plan
+- rename sre agent strings to argus agent
+- escape jsonpath braces in diagnose f-string prompt
+- improve diagnose prompt accuracy and filter low-score chunks from runbook --ask
+- add safe.directory config for PVC ownership mismatch
+- use deterministic MD5 point IDs in both vault ingest scripts
+- replace qdrant-client with httpx REST calls for search_vault
+- qdrant port 443, verify=False, batch upsert for vault ingestion
+- **ask**: add optional api key auth, fix default model name to mistral
+- **ask**: use LiteLLM OpenAI-compatible API instead of internal Ollama URL
+
+### Perf
+
+- parallelise status/diagnose/incident and add HTTP connection pooling
+- batch pod kubectl calls and parallelise flux sync checks
+- parallelise context gathering in sre ask

argus_sre-0.2.0/Dockerfile

@@ -0,0 +1,12 @@
+FROM python:3.12-slim
+
+WORKDIR /app
+
+COPY pyproject.toml README.md ./
+COPY argus ./argus
+
+RUN apt-get update && apt-get install -y --no-install-recommends git openssh-client && \
+    rm -rf /var/lib/apt/lists/* && \
+    pip install --no-cache-dir -e .
+
+CMD ["python", "-m", "argus.scripts.ingest_vault"]

argus_sre-0.2.0/PKG-INFO

@@ -0,0 +1,17 @@
+Metadata-Version: 2.4
+Name: argus-sre
+Version: 0.2.0
+Summary: Autonomous SRE intelligence for sovereign infrastructure
+License: Apache-2.0
+Requires-Python: >=3.11
+Requires-Dist: confluent-kafka>=2.3
+Requires-Dist: httpx>=0.27
+Requires-Dist: mcp[cli]>=1.0
+Requires-Dist: qdrant-client>=1.9
+Requires-Dist: rich>=13
+Requires-Dist: tomllib; python_version < '3.11'
+Requires-Dist: typer>=0.12
+Requires-Dist: watchdog>=4.0
+Provides-Extra: dev
+Requires-Dist: commitizen>=3.0; extra == 'dev'
+Requires-Dist: ruff>=0.4; extra == 'dev'

argus_sre-0.2.0/README.md

@@ -0,0 +1,148 @@
+# argus
+
+Autonomous SRE agent for Kubernetes — sovereign, on-premises, no data leaves the cluster.
+
+Built by a Lead SRE with 14 years of production infrastructure experience. Argus combines live cluster observability, semantic search over a local knowledge base, LLM-powered diagnosis, and an autonomous remediation agent with human-in-the-loop control.
+
+Every competitor (Datadog Bits AI, Dynatrace Davis, PagerDuty AI) sends your operational data to an external cloud. Argus doesn't.
+
+## Architecture
+
+```
+┌─────────────────────────────────────────────────┐
+│                    Argus                        │
+│                                                 │
+│  argus status    ──▶  Prometheus                │
+│  argus alert     ──▶  Alertmanager              │
+│  argus ask       ──▶  LiteLLM / Ollama          │
+│  argus diagnose  ──▶  kubectl + LLM             │
+│  argus runbook   ──▶  Qdrant (vector search)    │
+│  argus incident  ──▶  Qdrant + vault            │
+│  argus agent     ──▶  Alertmanager + kubectl    │
+│                       + LLM + Slack             │
+└─────────────────────────────────────────────────┘
+         │
+         ▼
+  Qdrant knowledge base
+  (runbooks, incident history, architecture docs)
+  ingested in real-time via Redpanda
+```
+
+## Commands
+
+```
+argus status                            Node health, firing alerts, high-restart pods
+argus alert                             Active Alertmanager alerts in plain English
+argus ask "<question>"                  LLM query with live Prometheus context
+argus diagnose <namespace/pod>          Logs + events + LLM root cause analysis
+argus diagnose <namespace/pod> --save   Diagnose and index result to Qdrant
+argus runbook <query>                   Semantic search over runbooks and vault notes
+argus runbook <query> --ask             Same, with LLM synthesis of top results
+argus incident start "<title>"          Open incident, snapshot cluster state
+argus incident note "<message>"         Append timestamped observation
+argus incident close "<summary>"        Close with resolution summary
+argus incident show                     Show active incident
+argus agent start                       Start autonomous daemon (human-in-the-loop)
+argus agent start --auto                Start in full-auto remediation mode
+argus agent approve                     Approve pending remediation action
+argus agent reject                      Reject pending remediation action
+argus agent skip                        Skip current alert
+argus agent status                      Show agent state and pending action
+```
+
+## Agent
+
+The agent daemon watches Alertmanager, diagnoses alerts automatically, and proposes remediation:
+
+```
+Alert fires → agent wakes → argus diagnose → search runbooks
+           → LLM picks action → notifies Slack → awaits approval
+           → executes → indexes outcome to Qdrant
+```
+
+Supported remediation actions:
+
+| Action | Trigger |
+|---|---|
+| `rollout_restart` | Pod crashlooping or stuck |
+| `rollout_undo` | New rollout caused the alert |
+| `force_delete_pod` | Pod stuck in Terminating or Unknown |
+| `scale_deployment` | Resource pressure |
+| `expand_pvc` | PVC near capacity |
+| `delete_failed_jobs` | Failed batch jobs accumulating |
+| `cordon_node` | Node unhealthy |
+| `drain_node` | Node needs evacuating |
+| `flux_reconcile` | Flux sync stuck or drifted |
+| `flux_suspend` | HelmRelease looping upgrade failures |
+| `restart_coredns` | DNS resolution failures cluster-wide |
+
+Human approval is required by default. Use `--auto` to execute immediately.
+
+## Installation
+
+```bash
+pip install argus-sre
+```
+
+Or from source (requires Python 3.11+ and [uv](https://github.com/astral-sh/uv)):
+
+```bash
+git clone https://github.com/t12-pybash/argus
+cd argus
+uv pip install -e .
+```
+
+## Configuration
+
+Config is loaded from `~/.config/argus/config.toml`. All values can be overridden with `ARGUS_` environment variables.
+
+```toml
+prometheus_url        = "https://prometheus.your-cluster.example"
+alertmanager_url      = "https://alertmanager.your-cluster.example"
+qdrant_url            = "https://qdrant.your-cluster.example"
+ollama_url            = "https://litellm.your-cluster.example"
+ollama_model          = "llama3.2"
+embed_ollama_url      = "http://localhost:11434"
+embed_model           = "nomic-embed-text"
+vault_path            = "/home/user/Documents/obsidian-vault"
+agent_webhook_url     = "https://hooks.slack.com/services/..."
+agent_approval_timeout = "300"
+```
+
+## Knowledge Base
+
+Argus maintains a Qdrant vector knowledge base of runbooks, incident history, and architecture docs. The `argus runbook` and `argus diagnose --save` commands search and write to this base, so the agent gets smarter with every incident.
+
+Real-time ingestion via Redpanda keeps the index in sync as vault documents are updated.
+
+## MCP Server
+
+Argus exposes an MCP server for use with Claude Code and other AI assistants:
+
+```bash
+claude mcp add argus -- argus-mcp
+```
+
+| Tool | Description |
+|---|---|
+| `get_cluster_status` | Node status, firing alerts, high-restart pods |
+| `get_alerts` | Active Alertmanager alerts with severity |
+| `prometheus_query` | Arbitrary PromQL execution |
+| `search_vault` | Semantic search over the Qdrant knowledge base |
+
+## Stack
+
+- [Typer](https://typer.tiangolo.com/) — CLI framework
+- [Rich](https://rich.readthedocs.io/) — terminal output
+- [FastMCP](https://github.com/jlowin/fastmcp) — MCP server
+- [httpx](https://www.python-httpx.org/) — HTTP client
+- [Qdrant](https://qdrant.tech/) — vector database
+- [Ollama](https://ollama.ai/) / [LiteLLM](https://github.com/BerriAI/litellm) — local LLM inference
+
+## Related
+
+- [homelab-gitops](https://github.com/t12-pybash/homelab-gitops) — the cluster this operates (Flux, Cilium, private AI platform)
+
+---
+
+*Ciarán Donegan — Lead SRE | [LinkedIn](https://www.linkedin.com/in/cdonegan7/) | [t-12.io](https://t-12.io)*

argus_sre-0.2.0/argus/clients/alertmanager.py

@@ -0,0 +1,11 @@
+import httpx
+
+
+class AlertmanagerClient:
+    def __init__(self, url: str):
+        self.url = url.rstrip("/")
+
+    def alerts(self) -> list:
+        r = httpx.get(f"{self.url}/api/v2/alerts", verify=False, timeout=10)
+        r.raise_for_status()
+        return r.json()