Remove em dashes and emojis across SDK

- Replace all em dashes with double hyphens
- Remove emojis from release_notes.md

Author: Prakashgode

CHANGELOG.md

@@ -1,12 +1,12 @@
 # Changelog
 
-## 0.0.20 — Real-Client MITM Validation
+## 0.0.20 -- Real-Client MITM Validation
 
 **Release date:** 2026-03-14
 
 This release marks the transition from "structurally validated" to "live MITM validated." The HTTPS interception path has been proven end-to-end with real HTTP clients through a trusted local CA.
 
-### HTTPS MITM — Now Real-Client Validated
+### HTTPS MITM -- Now Real-Client Validated
 
 - Validated with **curl**, **requests**, and **httpx** using a trusted local CA
 - 24 MITM live-validation tests covering: clean forward, blocked request, untrusted CA failure, secret in URL/header/body, binary passthrough, gzip body scanning, large body forwarding

release_notes.md

@@ -1,11 +1,11 @@
-#### **🚀 What's New?**
+#### **What's New**
 * **Verified Distribution:** Switched to OIDC (OpenID Connect) Trusted Publishing. Users can now verify the provenance of the package directly from our GitHub source.
 * **Project Metadata:** Added official links for documentation and issue tracking to improve developer support.
 * **Version Sync:** Standardized versioning across the SDK and PyPI.
 
-#### **🔒 Security Improvements**
+#### **Security Improvements**
 * **CI/CD Hardening:** Automated the release pipeline via GitHub Actions to ensure build integrity.
 * **Dependency Audit:** Verified `pyyaml` and `rich` requirements for the core scanner engine.
 
-#### **🛠 How to Update**
+#### **How to Update**
 `pip install --upgrade quanttape`

src/quanttape/proxy/README.md

@@ -1,4 +1,4 @@
-# QuantTape Guard — Proxy Module Technical Reference
+# QuantTape Guard -- Proxy Module Technical Reference
 
 ## Overview
 
@@ -82,7 +82,7 @@ When `cryptography` is not installed or cert generation fails:
 Client ◀──TCP──▶ _blind_tunnel() ◀──TCP──▶ Upstream
 ```
 
-Bidirectional byte relay with no inspection. Data passes through unmodified. This is the graceful degradation path — Guard still runs but HTTPS content is opaque.
+Bidirectional byte relay with no inspection. Data passes through unmodified. This is the graceful degradation path -- Guard still runs but HTTPS content is opaque.
 
 ---
 
@@ -97,7 +97,7 @@ scan_request(url, headers, body, rules=None, mode="agent") -> List[Finding]
 
 **How it differs from the scanner:**
 - No AST parsing (HTTP payloads aren't Python source)
-- No file I/O — operates on in-memory strings
+- No file I/O -- operates on in-memory strings
 - Scans URL, each header value, and each body line independently
 - Same regex patterns + Shannon entropy (threshold: 4.5, min length: 20 chars)
 - Sources are tagged as `url`, `header:{name}`, or `body:L{n}`
@@ -137,9 +137,9 @@ scan_request(url, headers, body, rules=None, mode="agent") -> List[Finding]
 ```
 
 **Stats tracking (in-memory):**
-- `requests_scanned` — total requests evaluated
-- `requests_blocked` — requests denied
-- `rules_triggered` — total individual rule hits
+- `requests_scanned` -- total requests evaluated
+- `requests_blocked` -- requests denied
+- `rules_triggered` -- total individual rule hits
 
 ---
 
@@ -189,8 +189,8 @@ The user must explicitly trust the CA cert. `quanttape setup-certs` prints platf
 - Minimal dependency footprint: only `httpx` for HTTP forwarding, `cryptography` for certs
 
 **Key classes:**
-- `GuardConfig` — dataclass: port, host, mode, config_path, no_verify
-- `GuardProxy` — main proxy handler with connection routing
+- `GuardConfig` -- dataclass: port, host, mode, config_path, no_verify
+- `GuardProxy` -- main proxy handler with connection routing
 
 **Methods:**
 | Method | Role |
@@ -216,8 +216,8 @@ Modes filter which rules are active:
 
 | Mode | Categories loaded | Use case |
 |---|---|---|
-| `agent` | `credential` | AI agent egress — catches secrets, skips trading logic |
-| `trading` | `credential`, `broker`, `trading_logic` | Trading bot egress — full broker key + logic checks |
+| `agent` | `credential` | AI agent egress -- catches secrets, skips trading logic |
+| `trading` | `credential`, `broker`, `trading_logic` | Trading bot egress -- full broker key + logic checks |
 | `all` | Everything | No filter, all rules active |
 
 Mode filtering happens in `rules.get_rules_for_mode()`. The bridge calls this when no explicit rule list is provided.
@@ -242,7 +242,7 @@ Mode filtering happens in `rules.get_rules_for_mode()`. The bridge calls this wh
 - Rules are compiled once per `scan_request()` call, not per line
 - Entropy check only runs on tokens >= 20 chars matching `[A-Za-z0-9/+=_-]{20,}`
 - HTTP forwarding uses httpx async client with connection pooling
-- Per-host certs are cached to disk — only generated once per hostname
+- Per-host certs are cached to disk -- only generated once per hostname
 - No disk I/O in the hot path (scan is pure in-memory regex)
 - **Measured latency (localhost, httpx client, 20 iterations per scenario):**
 

src/quanttape/proxy/server.py

@@ -106,7 +106,7 @@ async def _read_body(reader, content_length: int, headers: dict) -> tuple:
         return b"", None
 
     if not _is_text_content(headers):
-        return body_bytes, None  # binary content — skip scanning
+        return body_bytes, None  # binary content -- skip scanning
 
     body_text = _decode_body(body_bytes, headers)
     return body_bytes, body_text
@@ -325,7 +325,7 @@ async def _handle_connect(self, reader, writer, target):
         server_ctx.load_cert_chain(host_cert_path, host_key_path)
 
         # Upgrade the existing client connection to TLS using start_tls.
-        # Reuse the existing protocol/reader — start_tls reconnects them
+        # Reuse the existing protocol/reader -- start_tls reconnects them
         # to the new TLS transport automatically. This approach works
         # correctly on both Windows (ProactorEventLoop) and Unix.
         loop = asyncio.get_event_loop()

tests/benchmark_proxy.py

@@ -1,4 +1,4 @@
-"""QuantTape Guard — Proxy Benchmark Harness.
+"""QuantTape Guard -- Proxy Benchmark Harness.
 
 Measures latency overhead across dimensions:
   - Protocol: HTTP vs HTTPS (MITM)
@@ -305,7 +305,7 @@ async def cleanup():
 
     # ── Report ────────────────────────────────────────────────────
     print("\n" + "=" * 90)
-    print("  QUANTTAPE GUARD — BENCHMARK REPORT")
+    print("  QUANTTAPE GUARD -- BENCHMARK REPORT")
     print("=" * 90)
 
     print(f"\n  Baseline (direct HTTP, no proxy): "

tests/test_mitm_live.py

@@ -690,7 +690,7 @@ def print_matrix():
         return
 
     print("\n" + "=" * 80)
-    print("  QUANTTAPE GUARD — REAL-CLIENT MITM VALIDATION MATRIX")
+    print("  QUANTTAPE GUARD -- REAL-CLIENT MITM VALIDATION MATRIX")
     print("=" * 80)
 
     clients = {}

tests/test_proxy_e2e.py

@@ -252,7 +252,7 @@ def test_binary_content_passes_through(self):
                 body=binary_body,
             )
         )
-        # Should forward successfully (200 from upstream) — not blocked
+        # Should forward successfully (200 from upstream) -- not blocked
         self.assertEqual(status, 200)
 
     def test_gzip_body_scanned(self):