hermia 0.1.1__tar.gz

hermia-0.1.1/.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,49 @@
+---
+name: Bug report
+about: Something isn't working correctly
+title: 'bug: '
+labels: bug
+assignees: ''
+---
+
+## What happened
+
+<!-- Clear description of the bug. -->
+
+## Expected behavior
+
+<!-- What should have happened instead. -->
+
+## Steps to reproduce
+
+```
+1.
+2.
+3.
+```
+
+## Environment
+
+- OS:
+- Python version:
+- Ollama version:
+- Model(s) under test:
+- Hermia version / commit:
+
+## Eval context (if applicable)
+
+<!-- Which dimension or test ID is affected? Which model(s)? -->
+
+- Test ID:
+- Model:
+- Dimension:
+
+## Relevant output
+
+```
+paste terminal output here
+```
+
+## Additional context
+
+<!-- Anything else that might be relevant. -->

hermia-0.1.1/.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,36 @@
+---
+name: Feature request
+about: Propose a new eval test, dimension, or capability
+title: 'feat: '
+labels: enhancement
+assignees: ''
+---
+
+## What problem does this solve
+
+<!-- Why does this need to exist? What gap does it fill? -->
+
+## Proposed solution
+
+<!-- What would you build? Keep it specific. -->
+
+## Framework mapping (for new eval tests)
+
+<!-- If proposing a new eval test or dimension, which framework(s) does it map to?
+     OWASP LLM Top 10, MITRE ATLAS, CSA MAESTRO, NIST AI RMF — or a gap none of them cover. -->
+
+- Framework:
+- Reference (e.g. OWASP LLM01, ATLAS AML.T0054):
+
+## Alternatives considered
+
+<!-- What else could solve this problem? Why is your proposal better? -->
+
+## Scope estimate
+
+<!-- How much of the codebase does this touch?
+     New eval test only / schema changes / runner changes / TUI changes / CI changes -->
+
+## Additional context
+
+<!-- Links, prior art, related issues. -->

hermia-0.1.1/.github/dependabot.yml

@@ -0,0 +1,13 @@
+version: 2
+updates:
+  - package-ecosystem: pip
+    directory: "/"
+    schedule:
+      interval: weekly
+    open-pull-requests-limit: 5
+
+  - package-ecosystem: github-actions
+    directory: "/"
+    schedule:
+      interval: quarterly
+    open-pull-requests-limit: 5

hermia-0.1.1/.github/pull_request_template.md

@@ -0,0 +1,19 @@
+## What
+<!-- One sentence: what does this change do? -->
+
+## Why
+<!-- Why is this change needed? Link to issue or framework gap if applicable (e.g., OWASP LLM01, ATLAS AML.T0054). -->
+
+## Test coverage
+<!-- What tests cover this? New tests added? Existing tests affected? -->
+- [ ] New eval test cases added to `agentic-tasks.json`
+- [ ] Unit tests added/updated in `tests/`
+- [ ] Manually verified against local Ollama endpoint
+
+## Security checklist
+- [ ] No secrets, credentials, or API keys in diff
+- [ ] No new dependencies without justification
+- [ ] Bandit/pip-audit clean (CI will verify)
+
+## Notes for reviewer
+<!-- Anything Gemini Code Assist or a human reviewer should pay particular attention to. -->

hermia-0.1.1/.github/workflows/ci.yml

@@ -0,0 +1,34 @@
+name: CI
+
+on:
+  push:
+    branches:
+      - "feature/**"
+      - "fix/**"
+      - "chore/**"
+      - "dev"
+  pull_request:
+    branches:
+      - dev
+      - main
+
+jobs:
+  lint-and-test:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: pip install -e ".[dev]"
+
+      - name: Lint (ruff)
+        run: ruff check src/ tests/
+
+      - name: Type check (mypy)
+        run: mypy src/
+
+      - name: Test (pytest)
+        run: pytest tests/

hermia-0.1.1/.github/workflows/publish.yml

@@ -0,0 +1,55 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags:
+      - "v*.*.*"
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: pip install -e ".[dev]"
+
+      - name: Lint (ruff)
+        run: ruff check src/ tests/
+
+      - name: Type check (mypy)
+        run: mypy src/
+
+      - name: Test (pytest)
+        run: pytest tests/
+
+  build-and-publish:
+    needs: test
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      contents: read
+      id-token: write
+
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+
+      - name: Install build
+        run: pip install build==1.5.0
+
+      - name: Build
+        run: python -m build
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

hermia-0.1.1/.github/workflows/security.yml

@@ -0,0 +1,171 @@
+name: Security
+
+on:
+  pull_request:
+    branches:
+      - main        # full security gate before anything hits main
+      - dev         # catch issues before promotion, not after
+  schedule:
+    - cron: "0 6 * * 1"  # weekly Monday 06:00 UTC — scans main regardless of activity
+
+jobs:
+  gitleaks:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      pull-requests: read
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+        with:
+          fetch-depth: 0
+      - uses: gitleaks/gitleaks-action@ff98106e4c7b2bc287b24eaf42907196329070c7  # v2
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          GITLEAKS_CONFIG: .gitleaks.toml
+
+  trufflehog:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      pull-requests: read
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+        with:
+          fetch-depth: 0
+      - name: TruffleHog — entropy + credential scan
+        uses: trufflesecurity/trufflehog@0fa069c12f0c7baf431041cd1e564a9c5058846c  # main 2026-05-16
+
+  trivy:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+      - name: Trivy dependency scan
+        uses: aquasecurity/trivy-action@314ff8b43182423b84c50b1670b0e10f858f2d98  # master 2026-05-16
+        with:
+          scan-type: fs
+          scan-ref: "."
+          exit-code: "1"
+          severity: HIGH,CRITICAL
+
+  bandit:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+      - name: Install bandit
+        run: pip install bandit
+      - name: Bandit SAST
+        run: bandit -r src/ -ll
+
+  pip-audit:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+      - name: Audit hermia dependencies in isolated venv
+        run: |
+          python -m venv /tmp/audit-env
+          /tmp/audit-env/bin/pip install --quiet --upgrade pip
+          /tmp/audit-env/bin/pip install --quiet pip-audit
+          /tmp/audit-env/bin/pip install --quiet -e .
+          # CVE-2026-3219 affects pip itself (no fix available yet); all hermia deps are clean
+          /tmp/audit-env/bin/pip-audit --skip-editable --ignore-vuln CVE-2026-3219
+
+  guarddog-changes:
+    runs-on: ubuntu-latest
+    outputs:
+      relevant: ${{ steps.filter.outputs.relevant }}
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+        with:
+          fetch-depth: 0
+      - id: filter
+        run: |
+          if [[ "${{ github.event_name }}" == "schedule" ]]; then
+            echo "relevant=true" >> "$GITHUB_OUTPUT"
+          elif git diff --name-only "origin/${{ github.base_ref }}...HEAD" \
+               | grep -qE '^(pyproject\.toml|\.github/workflows/security\.yml)$'; then
+            echo "relevant=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "relevant=false" >> "$GITHUB_OUTPUT"
+            echo "No dep or workflow changes — skipping guarddog"
+          fi
+
+  guarddog:
+    needs: guarddog-changes
+    if: needs.guarddog-changes.outputs.relevant == 'true'
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405  # v6.2.0
+        with:
+          python-version: "3.11"
+      - name: Install guarddog
+        # pinned to avoid silent rule/CLI changes; bump quarterly and re-validate rule names
+        run: pip install "guarddog==2.10.0"
+      - name: GuardDog — typosquatting (hard fail)
+        run: |
+          python3 - <<'EOF'
+          import tomllib, subprocess, sys
+          from packaging.requirements import Requirement
+          from packaging.utils import canonicalize_name
+          with open("pyproject.toml", "rb") as f:
+              data = tomllib.load(f)
+          project = data.get("project", {})
+          all_deps = list(project.get("dependencies", []))
+          for group in project.get("optional-dependencies", {}).values():
+              all_deps += group
+          pkgs = sorted({canonicalize_name(Requirement(d).name) for d in all_deps})
+          failed = False
+          for pkg in pkgs:
+              print(f"Scanning {pkg}...")
+              r = subprocess.run(
+                  ["guarddog", "pypi", "scan", pkg,
+                   "--exit-non-zero-on-finding",
+                   "--rules", "typosquatting"],
+                  check=False,
+              )
+              if r.returncode != 0:
+                  failed = True
+          if failed:
+              sys.exit(1)
+          EOF
+      - name: GuardDog — heuristic rules (warn only)
+        continue-on-error: true
+        run: |
+          python3 - <<'EOF'
+          import tomllib, subprocess, os
+          from packaging.requirements import Requirement
+          from packaging.utils import canonicalize_name
+          with open("pyproject.toml", "rb") as f:
+              data = tomllib.load(f)
+          project = data.get("project", {})
+          all_deps = list(project.get("dependencies", []))
+          for group in project.get("optional-dependencies", {}).values():
+              all_deps += group
+          pkgs = sorted({canonicalize_name(Requirement(d).name) for d in all_deps})
+          findings = []
+          for pkg in pkgs:
+              print(f"Scanning {pkg}...")
+              r = subprocess.run(
+                  ["guarddog", "pypi", "scan", pkg,
+                   "--exit-non-zero-on-finding",
+                   "--rules", "potentially_compromised_email_domain",
+                   "--rules", "obfuscation"],
+                  check=False, capture_output=True, text=True,
+              )
+              print(r.stdout)
+              if r.returncode != 0:
+                  findings.append((pkg, r.stdout))
+          summary_path = os.environ.get("GITHUB_STEP_SUMMARY")
+          if summary_path and findings:
+              with open(summary_path, "a") as f:
+                  f.write("## :warning: GuardDog heuristic findings (warn only)\n\n")
+                  for pkg, output in findings:
+                      f.write(f"### `{pkg}`\n```\n{output.strip()}\n```\n\n")
+          EOF

hermia-0.1.1/.gitignore

@@ -0,0 +1,38 @@
+.venv/
+venv/
+__pycache__/
+._*
+.coverage
+uv.lock
+*.pyc
+*.pyo
+.mypy_cache/
+.ruff_cache/
+.pytest_cache/
+dist/
+*.egg-info/
+results/
+.env
+.env.*
+!.env.template
+session-notes/
+prompt.txt
+docs/talk-*.md
+docs/talk-*.js
+docs/*.pptx
+docs/*.pdf
+docs/slide-*.jpg
+docs/~$*
+logs/
+analysis/*.jsonl
+scripts/collect_and_push.sh
+scripts/run_analyze.sh
+scripts/seed_findings.py
+.claude/
+analysis/*.md
+analysis/*.html
+docs/*.docx
+
+# Secret scanning
+# .secrets.baseline is intentionally tracked (audited false-positive list)
+# .gitleaks.toml is intentionally tracked (custom rules)

hermia-0.1.1/.gitleaks.toml

@@ -0,0 +1,100 @@
+# Hermia — Gitleaks configuration
+# Extends the default ruleset with custom rules for internal infrastructure strings.
+# Run: gitleaks detect --config .gitleaks.toml
+# Docs: https://github.com/zricethezav/gitleaks/tree/master#configuration
+
+title = "hermia-gitleaks"
+
+[extend]
+# Inherit the full upstream default ruleset, then layer our rules on top.
+useDefault = true
+
+# ── Custom rules ─────────────────────────────────────────────────────────────
+
+[[rules]]
+id          = "tailscale-ip"
+description = "Tailscale CGNAT IP (100.64.0.0/10) — covers all fleet nodes past and future"
+regex       = '''(?i)(tailscale|ts[_\-]?ip|node[_\-]?ip|host)[^\n]{0,40}100\.(6[4-9]|[7-9][0-9]|1[01][0-9]|12[0-7])\.\d{1,3}\.\d{1,3}\b'''
+tags        = ["tailscale", "internal-ip"]
+
+[[rules]]
+id          = "tailscale-ip-bare"
+description = "Bare Tailscale CGNAT IP with no keyword context — belt-and-suspenders"
+regex       = '''\b100\.(6[4-9]|[7-9][0-9]|1[01][0-9]|12[0-7])\.\d{1,3}\.\d{1,3}\b'''
+tags        = ["tailscale", "internal-ip"]
+
+[[rules]]
+id          = "tailscale-hostname"
+description = "Any *.tailXXXX.ts.net hostname — catches all tailnets including future ones"
+regex       = '''\b[a-z0-9][a-z0-9\-]*\.tail[a-z0-9]+\.ts\.net\b'''
+tags        = ["tailscale", "internal-hostname"]
+
+[[rules]]
+id          = "rfc1918-172-16"
+description = "RFC 1918 172.16–31.x.x (Docker bridge, internal subnets)"
+regex       = '''\b172\.(1[6-9]|2[0-9]|3[01])\.\d{1,3}\.\d{1,3}\b'''
+tags        = ["internal-ip", "docker"]
+
+[[rules]]
+id          = "internal-ssh-alias"
+description = "Original internal SSH host aliases — catch if accidentally re-introduced"
+regex       = '''\b(m3pro-work|m3pro|m1pro-personal|m1pro|openclaw-linux|openclaw)\b'''
+tags        = ["internal-hostname", "ssh"]
+
+[[rules]]
+id          = "internal-path-home-scott"
+description = "/home/scott/* paths — personal home directory leakage"
+regex       = '''/home/scott/[^\s"'`]+'''
+tags        = ["internal-path"]
+
+[[rules]]
+id          = "grafana-dashboard-uid"
+description = "Grafana dashboard UID — full UUID or short UID (default 9-char) in URL or config"
+# Grafana native UIDs are typically 9-char alphanumeric (/d/<uid>/); full UUIDs also used
+regex       = '''grafana[^\n]{0,80}(/d/[A-Za-z0-9_-]{8,40}|[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12})'''
+tags        = ["grafana", "internal-id"]
+
+[[rules]]
+id          = "litellm-dispatch-key"
+description = "LiteLLM dispatch/master key"
+# Bridge pattern covers: env var assignments, Python dict literals, YAML config values
+regex       = '''(?i)(litellm[_\-]?(dispatch|master|api)[_\-]?key|LITELLM_[A-Z_]*KEY)[^\n]{0,40}["']?[A-Za-z0-9\-_]{20,}'''
+tags        = ["api-key", "litellm"]
+
+[[rules]]
+id          = "postgres-dsn"
+description = "PostgreSQL DSN with embedded credentials"
+# Known gap: passwords containing raw '@' are missed; %40 (URL-encoded) is matched correctly
+regex       = '''postgresql://[^:/?#\s]+:(?:[^@\s"'`\\]|%40){6,}@\S+'''
+tags        = ["database", "credentials"]
+
+# ── Allowlist ─────────────────────────────────────────────────────────────────
+
+[allowlist]
+description = "Safe strings that match rules but are intentionally present"
+regexes = [
+    # RFC 5737 documentation addresses — used as replacements after purge
+    '''192\.0\.2\.\d+''',
+    # RFC 2606 documentation domains
+    '''hermia\.example\.com''',
+    # Placeholder tokens in docs/tests
+    '''YOUR-DASHBOARD-ID''',
+    '''YOUR-TAILSCALE-IP''',
+    '''YOUR-TAILNET''',
+    # Literal regex patterns in this config file itself
+    '''100\\\.\(6\[4-9\]''',
+]
+
+paths = [
+    # Test fixtures that intentionally contain RFC-reserved example addresses
+    "tests/",
+    # Lock file — no secrets, just hashes
+    "uv.lock",
+    # detect-secrets baseline records false-positives as JSON — not real secrets
+    ".secrets.baseline",
+    # Fleet configs — intentionally contain fleet node hostnames and example labels
+    "hermia-fleet.yaml",
+    "fleet-.*\\.yaml",
+    "marcus-only.yaml",
+    "overnight.sh",
+]

hermia-0.1.1/.pre-commit-config.yaml

@@ -0,0 +1,18 @@
+repos:
+  - repo: https://github.com/gitleaks/gitleaks
+    rev: c2afd56bbe7ab11564107d34efac4bfde3e3c78d  # v8.24.0
+    hooks:
+      - id: gitleaks
+        args: ["--config", ".gitleaks.toml"]
+
+  - repo: https://github.com/Yelp/detect-secrets
+    rev: 01886c8a910c64595c47f186ca1ffc0b77fa5458  # v1.5.0
+    hooks:
+      - id: detect-secrets
+        args: ["--baseline", ".secrets.baseline"]
+        exclude: uv\.lock
+
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: cef0300fd0fc4d2a87a85fa2093c6b283ea36f4b  # v5.0.0
+    hooks:
+      - id: detect-private-key