feat: add private-line flagship surface

.github/FUNDING.yml

@@ -1,2 +1,4 @@
-github: HarperZ9
-custom: ["https://harperz9.github.io", "https://harperz9.github.io/field-guide.html"]
+github: [HarperZ9]
+custom:
+  - https://harperz9.github.io
+  - https://harperz9.github.io/field-guide.html

.github/workflows/ci.yml

@@ -0,0 +1,32 @@
+name: behavior-transform.io CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+
+      - name: Setup Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.11"
+
+      - name: Install test dependencies
+        run: python -m pip install pytest
+
+      - name: Run tests
+        run: python -m pytest -q
+
+      - name: Verify flagship envelope
+        run: |
+          python -m unittest tests.test_behavior_flagship -v
+          python tools/behavior_flagship.py doctor --json

README.md

@@ -0,0 +1,63 @@
+# behavior-transform.io
+
+![behavior-transform.io IO boundary calibration layer](assets/behavior-transform-hero.svg)
+
+`behavior-transform.io` is the Project Telos private-line IO boundary
+calibration layer. It provides local wrappers, hooks, and shell profiles for
+read, write, exec, fetch, input, and model-boundary channels.
+
+Its job is practical: make workstation IO explicit, mode-aware, and receipt
+friendly without requiring hosts to ingest raw prompts, private file contents,
+secret values, or full model-boundary payloads.
+
+```bash
+python tools/behavior_flagship.py status --json
+python tools/behavior_flagship.py doctor --json
+python tools/behavior_flagship.py demo --json
+```
+
+## Flagship Contract
+
+| Surface | Status |
+|---------|--------|
+| CLI JSON | `tools/behavior_flagship.py status|doctor|demo --json` |
+| Runtime profiles | `ops`, `research`, `academic` |
+| Hook surface | Claude Code PreToolUse/PostToolUse hooks in `hooks/` |
+| Shell surface | PowerShell, CMD, and sh profile helpers in `profiles/` |
+| Interop schemas | `project-telos.flagship-action/v1`, `project-telos.action-receipt/v1`, `project-telos.context-envelope/v1` |
+| Privacy boundary | Host receives counts, hashes, verdicts, and redacted refs; local adapters retain raw content |
+
+## Modes
+
+| Mode | Profile | Behavior |
+|------|---------|----------|
+| `ops` | standard | Calibration stack active |
+| `research` | native | Source-faithful passthrough |
+| `academic` | native | Research alias |
+
+Switch mode:
+
+```bash
+python tools/io_state.py --set research
+python tools/io_state.py --set ops
+```
+
+## Structure
+
+```text
+tools/      Core modules and CLI wrappers
+hooks/      Claude Code hook adapters
+profiles/   Shell integration helpers
+tests/      Pytest and unittest coverage
+docs/       Specs, plans, and integration contracts
+```
+
+## Verification
+
+```bash
+python -m pytest -q
+python tools/behavior_flagship.py doctor --json
+```
+
+See [docs/INTEGRATION_CONTRACT.md](docs/INTEGRATION_CONTRACT.md) for the IO
+boundary contract.

docs/INTEGRATION_CONTRACT.md

@@ -0,0 +1,62 @@
+# behavior-transform.io Integration Contract
+
+`behavior-transform.io` is the private-line IO boundary calibration layer for
+Project Telos. It provides local wrappers, hooks, and shell profiles for moving
+text through read, write, exec, fetch, input, and model-boundary channels with
+explicit mode and provenance handling.
+
+This repository is not an offensive tooling layer and does not own target
+operations. It owns the workstation-local IO membrane around tools that do.
+
+## Runtime Profiles
+
+| Profile | Purpose | Transform behavior |
+|---------|---------|--------------------|
+| `ops` | Default operator mode | Calibration enabled |
+| `research` | Source-faithful research mode | Passthrough |
+| `academic` | Alias for research review | Passthrough |
+
+Switching profile is local state only:
+
+```powershell
+python tools/io_state.py --set research
+python tools/io_state.py --set ops
+```
+
+## Host Surfaces
+
+- CLI JSON envelopes through `tools/behavior_flagship.py`
+- Claude Code hooks in `hooks/`
+- shell profiles in `profiles/`
+- direct safe wrappers in `tools/safe_*.py`
+- future MCP adapter surface using the same envelope schema
+
+## Privacy Boundary
+
+Hosts may receive:
+
+- mode and profile names
+- wrapper/tool names
+- counts
+- hashes
+- verdicts
+- redacted references
+
+Hosts must not require raw prompts, private file contents, secret values,
+credentials, hidden environment values, or full model-boundary payloads.
+
+## Receipt Shape
+
+Actions that cross write, exec, fetch, or model-boundary channels should record:
+
+- `action_intent_id`
+- `channel`
+- `profile`
+- `input_hash`
+- `output_hash` or `redacted_output_ref`
+- `decision_outcome`
+- `verification_verdict`
+- `evaluated_at`
+
+The receipt must be sufficient to replay the boundary decision without exporting
+private source content.

tests/test_behavior_flagship.py

@@ -0,0 +1,43 @@
+import unittest
+
+from tools.behavior_flagship import (
+    SCHEMA,
+    demo_envelope,
+    doctor_envelope,
+    status_envelope,
+)
+
+
+class BehaviorFlagshipTests(unittest.TestCase):
+    def test_status_uses_telos_schema(self) -> None:
+        envelope = status_envelope()
+
+        self.assertEqual(envelope["schema"], SCHEMA)
+        self.assertEqual(envelope["tool"], "behavior-transform.io")
+        self.assertEqual(envelope["status"], "MATCH")
+        self.assertIn("CLI JSON", envelope["native"]["host_surfaces"])
+        self.assertFalse(envelope["native"]["runtime_contract"]["raw_secret_export"])
+
+    def test_doctor_checks_contract_and_secret_boundary(self) -> None:
+        envelope = doctor_envelope()
+        checks = {
+            item["id"]
+            for output in envelope["outputs"]
+            if output["kind"] == "checks"
+            for item in output["items"]
+        }
+
+        self.assertIn("required:docs/INTEGRATION_CONTRACT.md", checks)
+        self.assertIn("raw-secret-boundary", checks)
+        self.assertEqual(envelope["status"], "MATCH")
+
+    def test_demo_is_local_io_only(self) -> None:
+        envelope = demo_envelope()
+        demo = envelope["outputs"][0]
+
+        self.assertEqual(demo["runtime_surface"], "local_io")
+        self.assertIn("switch back to ops calibration", demo["steps"])
+
+
+if __name__ == "__main__":
+    unittest.main()