crprotocol 5.0.0__tar.gz → 5.1.0__tar.gz

{crprotocol-5.0.0 → crprotocol-5.1.0}/.github/workflows/ci.yml

@@ -30,7 +30,7 @@ jobs:
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          pip install -e ".[dev]"
+          pip install -e ".[full,mcp,dev]"
 
       - name: Lint with ruff
         run: ruff check crp/ tests/

{crprotocol-5.0.0 → crprotocol-5.1.0}/.gitignore

@@ -51,6 +51,8 @@ encrypted-*.pem
 .env.staging
 .env.development
 secrets/
+kimi_moonshot_api_key.txt
+*_api_key.txt
 
 # Dev scratch files (root-level only)
 /_*.py

{crprotocol-5.0.0 → crprotocol-5.1.0}/AI Agent adoption/CRP-SKILL.md

@@ -1,7 +1,7 @@
 # CRP — Context Relay Protocol: Agent Integration Guide
 
 > This file teaches an AI coding agent how to add AI governance, safety, and compliance
-> to an application using CRPv4. If you are an AI agent helping a developer govern, secure,
+> to an application using CRPv5. If you are an AI agent helping a developer govern, secure,
 > or make their LLM application audit-ready, this is your reference.
 >
 > For the short discovery file, see `https://crprotocol.io/llms.txt`.

crprotocol-5.0.0/site-docs/llms-full.txt → crprotocol-5.1.0/AI Agent adoption/llms-full.txt

@@ -1,7 +1,7 @@
 # CRP — Context Relay Protocol: Agent Integration Guide
 
 > This file teaches an AI coding agent how to add AI governance, safety, and compliance
-> to an application using CRPv4. If you are an AI agent helping a developer govern, secure,
+> to an application using CRPv5. If you are an AI agent helping a developer govern, secure,
 > or make their LLM application audit-ready, this is your reference.
 >
 > For the short discovery file, see `https://crprotocol.io/llms.txt`.

{crprotocol-5.0.0 → crprotocol-5.1.0}/AI Agent adoption/llms.txt

@@ -1,4 +1,4 @@
-# CRP — Context Relay Protocol (v4)
+# CRP — Context Relay Protocol (v5)
 
 > CRP is a sidecar governance protocol for AI inference. It sits between an application and
 > any LLM provider and governs every call: enforcing grounding and safety policy, scoring

{crprotocol-5.0.0 → crprotocol-5.1.0}/CHANGELOG.md

@@ -11,7 +11,87 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 
 ## [Unreleased]
 
-## [5.0.0] - 2026-06-30
+## [5.1.0] - 2026-07-04
+
+### Fixed
+- **Context-overflow guard in the positioned loop (`crp/stl/positioned.py`).**
+  `provider_model_call` previously sent a fixed `max_tokens` with no check against
+  the connected provider's real context window, even though `count_tokens()` and
+  `context_window_size()` are mandatory on the `LLMProvider` ABC. Added
+  `guard_prompt_budget()` (exported via `crp.stl`): caps `max_tokens` and, only if
+  still necessary, trims the **oldest** prompt content (preserving the task/operation
+  frame) so input, tool-call frames, continuation windows, and accumulated
+  multi-turn CSO state can never overflow the model's actual window — local or
+  frontier. Verified end-to-end against a live 8196-token local model across a
+  6-turn multi-turn/tool-call/continuation stress scenario with zero overflow
+  errors (see `crp-comply/scripts/context_overflow_stress.py`).
+- **Continuation anti-repetition** (`crp/stl/positioned.py`): the continuation loop
+  now feeds a `DocumentMap`-derived table of covered sections into each window's
+  prompt and applies an n-gram repetition guard (retry once, then stop) — reduces
+  4-gram repetition on Kimi to a 3.66% mean across the SQB test cases (was
+  15–30% under the naive continuation harness).
+- **GitHub App: wrong installation-repos endpoint** (`crp/scan/github_app.py`).
+  `list_repos()` called `GET /installations/{installation_id}/repositories`,
+  which is not a real GitHub API route (always 404, regardless of a valid
+  installation ID/token). The correct endpoint is `GET /installation/repositories`
+  (singular, no ID in the path — the installation is implied entirely by the
+  installation access token used for auth). Also switched the auth header to
+  `Bearer` (matching `X-GitHub-Api-Version: 2022-11-28`) for consistency with
+  `installation_token()`. Added a URL-asserting regression test — the previous
+  test mocked `requests.get` without checking the actual URL, so it never
+  caught this.
+- **`crp.scan.remediation.RemediationEngine.open_pr()` crashed in production**
+  whenever `crp.scan.github_app` was importable (i.e. always, once installed):
+  it called `GithubAppClient()` with no arguments, but the class requires
+  `app_id` + `private_key_pem`. Fixed to use the existing
+  `GithubAppClient.from_env()` factory, catching `RuntimeError` (env vars not
+  configured) alongside the existing `ImportError` (module not installed) —
+  both now return the same graceful fallback string instead of crashing.
+- **`crp_mcp.auth.ClerkTokenVerifier` / `authenticate()` used non-existent
+  `AccessToken` fields.** The real `mcp.server.auth.provider.AccessToken`
+  pydantic model only has `token`/`client_id`/`scopes`/`expires_at`/`resource`
+  — no `subject` or `claims` field. Constructing it with those kwargs silently
+  dropped them, and `authenticate()` reading `access_token.claims` would raise
+  `AttributeError` in production whenever FastMCP's auth middleware path was
+  taken (not caught, not covered by any prior test). Fixed to store the
+  resolved user id in `client_id` and re-derive claims via
+  `_verify_token(access_token.token)` (the JWKS client is already cached).
+  Added a regression test exercising the FastMCP-context path directly.
+- **`.github/workflows/ci.yml` never installed the `[mcp]` extras** (mcp,
+  pyyaml, httpx, `pyjwt[crypto]`, stripe, google-auth) — only `[dev]`. This
+  meant `crp_mcp`/comply/scan tests that depend on those packages could never
+  correctly run in CI. Changed to `pip install -e ".[full,mcp,dev]"`.
+- **Windows GLiNER/PyTorch native crash (fatal access violation during
+  `GLiNER.from_pretrained()`).** Root-caused as a Windows OpenMP
+  duplicate-runtime conflict — numpy, scipy, scikit-learn, and torch each
+  bundle their own OpenMP DLL, and loading more than one in the same process
+  can crash natively when GLiNER's model weights are loaded. Not a CRP logic
+  bug, but genuinely fixable rather than merely a "disable stage 3"
+  workaround: `crp/__init__.py` now sets `KMP_DUPLICATE_LIB_OK=TRUE` and
+  `OMP_NUM_THREADS=1` (Windows-only, before any other CRP import, so it takes
+  effect before numpy/scipy/torch are first loaded), and
+  `crp/extraction/stage3_gliner.py` additionally caps `torch.set_num_threads(1)`
+  at the GLiNER model-load call site. Verified via a repeated fresh-load
+  stress test (15/15 clean runs) with zero effect on Linux/macOS.
+- Several previously-failing tests were stale rather than indicating real
+  bugs, and have been corrected to match intentional prior refactors:
+  `test_gap_fixes_live.py` (7 tests) checked `CRPOrchestrator.dispatch`
+  directly via `inspect.getsource()`, but `dispatch()` is now a thin
+  thread-safety wrapper around `_dispatch_locked()` (added for reentrant
+  locking) — updated to inspect `_dispatch_locked`; two more assumed eager
+  embedding-model loading, which is intentionally lazy by default
+  (`eager_load_models`) — updated to opt in explicitly.
+  `test_comply_quota_gate.py` (2 tests) hardcoded a stale `"2026-06"` billing
+  period key that no longer matches `period_key()`'s real current-month
+  output — updated to call `period_key()`. `test_orchestrator_perf.py`'s
+  Ollama-probe-timeout test assumed no local Ollama instance is ever running —
+  now mocks `urllib.request.urlopen` so it is deterministic regardless of the
+  host machine. `test_sdk_generated_reference.py` (2 tests) failed because
+  `site-docs/api/modules/` was out of sync with the current module list
+  (`crp.tools` was added without regenerating docs) — regenerated via
+  `scripts/generate_api_reference.py`.
+
+
 
 CRP v5 turns the protocol from an inference-governance layer into a full
 **agentic execution** layer for small/local language models. The doctrine is

{crprotocol-5.0.0 → crprotocol-5.1.0}/CRPv4_Implementation_Rounds.md

@@ -5,7 +5,10 @@
 **Target:** CRP v4.0.0  
 **PyPI publish token (use only after v4.0.0 fully passes SQB gate):**
 ```
-pypi-AgEIcHlwaS5vcmcCJDZhOTJhZWMyLWU1NGUtNGI5MS05NTQ1LTllNTkxZGZhMTM2ZgACElsxLFsiY3Jwcm90b2NvbCJdXQACLFsyLFsiMTcwNjA3MTAtMzQwNC00YTM3LWEyNmYtYzgwNzY5YzQ4NjA3Il1dAAAGIDFak4jg6Ainb5mLvB5soBvA7edkdk7mYjDUZ-awLtTe
+[REDACTED 2026-07-02 -- this token was committed in plaintext and has been
+ removed. Treat it as compromised; rotate on PyPI if not already rotated.
+ NEVER commit publish tokens to the repository -- pass via TWINE_PASSWORD
+ env var or an untracked local file instead.]
 ```
 
 ---
@@ -169,7 +172,7 @@ git push origin v4.0.0
 
 #### 2e. PyPI publish
 ```cmd
-.venv\Scripts\python.exe -m twine upload dist/crprotocol-4.0.0-*.whl --username __token__ --password pypi-AgEIcHlwaS5vcmcCJDZhOTJhZWMyLWU1NGUtNGI5MS05NTQ1LTllNTkxZGZhMTM2ZgACElsxLFsiY3Jwcm90b2NvbCJdXQACLFsyLFsiMTcwNjA3MTAtMzQwNC00YTM3LWEyNmYtYzgwNzY5YzQ4NjA3Il1dAAAGIDFak4jg6Ainb5mLvB5soBvA7edkdk7mYjDUZ-awLtTe
+.venv\Scripts\python.exe -m twine upload dist/crprotocol-4.0.0-*.whl --username __token__ --password %PYPI_TOKEN%
 ```
 
 ### PYTEST NOTE (for B4.1)
@@ -1385,7 +1388,7 @@ Document all v4.0.0 changes:
 pip install twine
 python -m twine upload dist/crprotocol-4.0.0-*.whl \
   --username __token__ \
-  --password pypi-AgEIcHlwaS5vcmcCJDZhOTJhZWMyLWU1NGUtNGI5MS05NTQ1LTllNTkxZGZhMTM2ZgACElsxLFsiY3Jwcm90b2NvbCJdXQACLFsyLFsiMTcwNjA3MTAtMzQwNC00YTM3LWEyNmYtYzgwNzY5YzQ4NjA3Il1dAAAGIDFak4jg6Ainb5mLvB5soBvA7edkdk7mYjDUZ-awLtTe
+  --password "$PYPI_TOKEN"
 ```
 
 #### B4.4 — GitHub Push

{crprotocol-5.0.0 → crprotocol-5.1.0}/CRPv5_FINALISATION_TRACKING.md

@@ -49,8 +49,8 @@ Legend: ☐ not started · ◐ in progress · ☑ done
 | **4** | **Make STL live** — real positioned loop (`run_positioned`); preventive-safety gate; INTEGRATE→CSO; structured-decoding contract at the `model_call` boundary | `crp/stl/positioned.py` | ☑ **done** — live loop + 8B validation; full 13-stage DPE verify is a follow-up |
 | **5** | **Observation Store + Resource Governor + Visibility event stream** | `crp/state/`, `crp/observability/`, `crp/resources/` | ☑ **done** — `ResourceGovernor` (slow-and-steady, device-tier capping); observations→CSO; OSM event stream |
 | **6** | **CLARIFY + Checkpoint consolidation** — make checkpoints faultless, test them, wire CLARIFY | `crp/security/clarify.py`, `crp/security/checkpoint.py` | ☑ **done** — sync `clarify` bridge (Invariant 10 fallback); CLARIFY op + oversight approval wired; async `Checkpoint` retained for webhooks |
-| **7** | **Expose** — `client.dispatch_positioned()` / execution-mode `positioned-tool-loop`; preventive safety gate; SDK + Gateway wiring | `crp/sdk/client.py`, `crp/core/orchestrator.py`, `crp-gateway/` | ☐ |
-| **8** | **Version bump 5.0.0**, promote SPEC-049/050 from draft, conformance, CHANGELOG | `crp/_version.py`, specs, `CHANGELOG.md` | ☐ |
+| **7** | **Expose** — `client.dispatch_positioned()` / execution-mode `positioned-tool-loop`; preventive safety gate; SDK + Gateway wiring | `crp/sdk/client.py`, `crp/core/orchestrator.py`, `crp-gateway/` | ☑ **done** — `dispatch_positioned()` wired; `CRP-Agent-*` headers; validated live on 8B (191-token bounded window). Gateway-product wiring tracked separately in the `crp-gateway` repo |
+| **8** | **Version bump 5.0.0**, promote SPEC-049/050 from draft, conformance, CHANGELOG | `crp/_version.py`, specs, `CHANGELOG.md` | ☑ **done** — 5.0.0 shipped to PyPI; SPEC-049/050 → Stable; CHANGELOG written; tagged `v5.0.0` |
 
 ---
 
@@ -161,5 +161,40 @@ CRP makes **two distinct guarantees**, and it is important not to conflate them:
 
 ---
 
-*Tracking report generated 2026-06-30. Update the status columns in §2 and the gate
-checklist in §6 as work lands.*
+## 9. Realised state audit (2026-07-01) — the honest answer
+
+> **What is really the state of CRPv5?** The positioned agentic loop is **real and
+> working for single- and multi-turn requests**, proven end-to-end on Kimi (frontier)
+> and the local 8B. It is **not yet** wired to the v4 deep-context/continuation
+> subsystems. Below is the audited truth.
+
+### Working (verified end-to-end, local 8B + Kimi + logic)
+| Use case | Status | Evidence |
+|----------|--------|----------|
+| Tool-call agentic execution | ✅ working | `run_positioned` + TCF + executor; e2e local 3/3, Kimi 3/3 |
+| Context positioning (in-turn) | ✅ working | CSO `to_prompt_context` carries facts across operations; e2e PASS |
+| **Multi-turn state relay** | ✅ **NEW, working** | `prior_cso` param seeds established_facts/observations/decisions; window advances; e2e Kimi PASS + regression test `test_multi_turn_state_relay` |
+| CLARIFY / human-in-the-loop | ✅ working | `resolve_clarification`; e2e PASS |
+| Preventive oversight halt | ✅ working | `_preventive_check` + oversight set; e2e PASS |
+| Resource-governed profile | ✅ working | `ResourceGovernor.plan`; e2e PASS |
+| Bounded working set | ✅ working | 68 tok across 5 relayed turns; ≤206 tok on real tasks |
+
+### Honest gaps (NOT yet wired into the positioned loop)
+| Gap | Impact | Where it lives |
+|-----|--------|----------------|
+| Output continuation (one operation exceeding the token wall) | Long single-operation generations truncate; mitigated because work is split into operations and turns | `crp/continuation/` exists, not called from `positioned.py` |
+| CDR/CDGR auto-retrieval | `context_facts` is passed in manually, not retrieved by novelty/graph walk | `crp/envelope/cdr.py`, `crp/ckf/cdgr.py` exist, not imported |
+| Multi-horizon context / scratch buffer | Frame builder uses a flat fact list, no PERSISTENT/CONVERSATIONAL/EPHEMERAL blend | `crp/state/horizons.py`, `scratch_buffer.py` exist, not imported |
+| DPE verification in the loop | No credibility/entailment gate on each operation's output | `crp/provenance/` exists, not imported |
+
+### Verdict
+The loop is a **real agentic executor with proven multi-turn continuity and bounded
+windows**. The remaining gaps are *depth* enhancements (auto-retrieval, output
+continuation, DPE gating) — valuable next, but the core "positioning + multi-turn +
+context carry-forward" is done and tested. Test harnesses: `examples/crp_demos/e2e_v5_test.py`
+(every use case, local+Kimi+logic) and `examples/crp_demos/positioned_benchmark.py`.
+
+---
+
+*Tracking report generated 2026-06-30; realised-state audit added 2026-07-01.*
+

{crprotocol-5.0.0 → crprotocol-5.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: crprotocol
-Version: 5.0.0
+Version: 5.1.0
 Summary: Context Relay Protocol — unbounded context, unbounded generation, amplified reasoning for LLMs
 Project-URL: Homepage, https://crprotocol.io
 Project-URL: Documentation, https://crprotocol.io
@@ -89,13 +89,13 @@ Description-Content-Type: text/markdown
 <p align="center">
   <a href="LICENSE.md"><img src="https://img.shields.io/badge/Spec-CC_BY--SA_4.0-blue.svg" alt="Spec: CC BY-SA 4.0"></a>
   <a href="LICENSE.md"><img src="https://img.shields.io/badge/Code-ELv2-orange.svg" alt="Code: Elastic License 2.0"></a>
-  <img src="https://img.shields.io/badge/Spec_Version-4.0.0-brightgreen.svg" alt="Spec Version: 4.0.0">
+  <img src="https://img.shields.io/badge/Spec_Version-5.1.0-brightgreen.svg" alt="Spec Version: 5.1.0">
   <img src="https://img.shields.io/badge/RFC_2119-Conformant-orange.svg" alt="RFC 2119">
   <img src="https://img.shields.io/badge/Language_Neutral-JSON_Schema-yellow.svg" alt="Language Neutral">
-  <img src="https://img.shields.io/badge/Status-v4.0.0_WIP-yellow.svg" alt="Status: v4.0.0 WIP">
+  <img src="https://img.shields.io/badge/Status-v5.1.0_Stable-brightgreen.svg" alt="Status: v5.1.0 Stable">
   <a href="https://github.com/AutoCyber-AI/context-relay-protocol/actions"><img src="https://img.shields.io/github/actions/workflow/status/AutoCyber-AI/context-relay-protocol/ci.yml?label=CI" alt="CI"></a>
   <img src="https://img.shields.io/badge/python-3.10%2B-blue.svg" alt="Python 3.10+">
-  <img src="https://img.shields.io/badge/tests-1%2C537-brightgreen.svg" alt="1,537 tests">
+  <img src="https://img.shields.io/badge/tests-2%2C796-brightgreen.svg" alt="2,796 tests">
 </p>
 
 <p align="center">
@@ -1136,7 +1136,7 @@ Full error taxonomy with all codes in [§6.10.4](specification/02_CORE_PROTOCOL.
 
 | Language | Status | Package | Repository |
 |----------|--------|---------|------------|
-| **Python** | ✅ v3.1.1 / v4.0.0 WIP | `pip install -e ".[dev]"` | This repository |
+| **Python** | ✅ v5.1.0 | `pip install crprotocol` | This repository |
 | **TypeScript** | 📋 Planned | `npm install @crp/sdk` | `crp-typescript` |
 | **Rust** | 📋 Planned | `cargo add crp` | `crp-rust` |
 
@@ -1270,11 +1270,13 @@ All configuration is optional. CRP works with zero configuration if you pass you
 ## Roadmap
 
 ### Phase 1: Open Specification ✅
-### Phase 4: v4.0 Implementation ← **We are here**
+### Phase 5: v5.1 Agentic Execution ← **We are here**
 - [x] Publish CRP v2.0 specification (9 documents, ~19,200 lines)
 - [x] JSON Schema definitions for all API types
-- [ ] Reference SDK: Python (`pip install crprotocol`)
-- [ ] Benchmark results: CRP on vs. off across tasks and models
+- [x] Reference SDK: Python (`pip install crprotocol`)
+- [x] Positioned agentic loop + Tool Capability Fabric (SPEC-049/050)
+- [x] Multi-turn CSO relay and continuation anti-repetition
+- [ ] Benchmark results published to BENCHMARKS.md
 - [ ] arXiv technical report with empirical evaluation
 
 ### Phase 2: Ecosystem
@@ -1398,7 +1400,7 @@ See [LICENSE.md](LICENSE.md) for the full license text.
 ---
 
 <p align="center">
-  <strong>Context Relay Protocol v4.0.0 WIP</strong><br>
+  <strong>Context Relay Protocol v5.1.0</strong><br>
   Zero configuration. Unbounded input. Unbounded output. Amplified reasoning.<br>
   Better context at every scale. Honest degradation. Quality-tiered.<br>
   Peak quality. Every window.

{crprotocol-5.0.0 → crprotocol-5.1.0}/RAILWAY_DEPLOYMENT_GUIDE.md

@@ -58,11 +58,13 @@ In the Railway project → click your service → **Variables** tab → **"New V
 ```
 # ── Clerk ────────────────────────────────────────────────────────
 CLERK_ISSUER=https://clerk.scribe.crprotocol.io
-CLERK_SECRET_KEY=sk_live_j5y6TVVydxWe93qEj9DAUhB9jZJtpM8i4Ti4mVi5XQ
+CLERK_SECRET_KEY=sk_live_...
 
-# ── Stripe ────────────────────────────────────────────────────────
+# ── Stripe ──────────────────────────────────────────────────────────
+# NEVER commit real values here. Pull from your local .env / secrets manager
+# and paste directly into Railway's Variables UI -- never into this file.
 STRIPE_SECRET_KEY=<value from your .env file>
-STRIPE_WEBHOOK_SECRET=whsec_ASew8AShDg6NtD4jn4F6O1lY3FYv4SDz
+STRIPE_WEBHOOK_SECRET=whsec_...
 STRIPE_SCRIBE_PRO_PRICE_ID=price_1TMSB5GRBK524I7zhmUgLmfm
 STRIPE_SCRIBE_ENTERPRISE_PRICE_ID=price_1TMSB5GRBK524I7zCIU71YN6
 
@@ -92,11 +94,11 @@ VITE_CLERK_PUBLISHABLE_KEY=pk_live_<get-from-clerk-dashboard>
 ```
 # ── Clerk ────────────────────────────────────────────────────────
 CLERK_ISSUER=https://clerk.comply.crprotocol.io
-CLERK_SECRET_KEY=sk_live_M96E3pt7KwY3WnsIERdW7VHUbZ1NfLLnExvCyq3Qom
+CLERK_SECRET_KEY=sk_live_...
 
-# ── Stripe ────────────────────────────────────────────────────────
+# ── Stripe ──────────────────────────────────────────────────────────
 STRIPE_SECRET_KEY=<value from your .env file>
-STRIPE_WEBHOOK_SECRET=whsec_TzhIRl0XKlqfRVGlVUOI6MwZmzlLxhum
+STRIPE_WEBHOOK_SECRET=whsec_...
 STRIPE_COMPLY_PRO_PRICE_ID=price_1TMSGAGRBK524I7zvSYEwg9S
 STRIPE_COMPLY_ENTERPRISE_PRICE_ID=price_1TMSGAGRBK524I7z9ebppShK
 STRIPE_COMPLY_CLOUD_PRICE_ID=price_1TMSGAGRBK524I7zaSuQMMzb

{crprotocol-5.0.0 → crprotocol-5.1.0}/RAILWAY_VARIABLES.md

@@ -20,12 +20,12 @@ Paste these directly into Railway → Service → Variables → **Raw Editor**.
 
 ```
 CLERK_ISSUER=https://clerk.scribe.crprotocol.io
-CLERK_SECRET_KEY=sk_live_j5y6TVVydxWe93qEj9DAUhB9jZJtpM8i4Ti4mVi5XQ
-STRIPE_SECRET_KEY=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
-STRIPE_WEBHOOK_SECRET=whsec_ASew8AShDg6NtD4jn4F6O1lY3FYv4SDz
+CLERK_SECRET_KEY=sk_live_...
+STRIPE_SECRET_KEY=sk_live_...
+STRIPE_WEBHOOK_SECRET=whsec_...
 STRIPE_SCRIBE_PRO_PRICE_ID=price_1TMSB5GRBK524I7zhmUgLmfm
 STRIPE_SCRIBE_ENTERPRISE_PRICE_ID=price_1TMSB5GRBK524I7zCIU71YN6
-CRP_SCRIBE_JWT_SECRET=81ab043b53aece66d5347efe2de858b8f65817f8a740ac9ec386e86c058f0000
+CRP_SCRIBE_JWT_SECRET=<generate-your-own-random-secret>
 CRP_SCRIBE_DATA_DIR=/app/data
 CRP_SCRIBE_FRONTEND_DIR=/app/frontend/dist
 CRP_SCRIBE_BASE_URL=https://scribe.crprotocol.io
@@ -33,18 +33,23 @@ CRP_SCRIBE_CORS_ORIGINS=https://scribe.crprotocol.io
 PYTHONUNBUFFERED=1
 ```
 
+> **NEVER commit real secret values to this file.** Pull the actual values from
+> your local `.env` / secrets manager and paste them directly into Railway's
+> Variables UI -- never into a git-tracked file. This file documents variable
+> *names* and *shape* only.
+
 ### All Scribe Variables at a Glance
 
 | Variable | Value | Build-time? |
 |---|---|---|
 | `VITE_CLERK_PUBLISHABLE_KEY` | `pk_live_Y2xlcmsuc2NyaWJlLmNycHJvdG9jb2wuaW8k` | **YES — toggle ON** |
 | `CLERK_ISSUER` | `https://clerk.scribe.crprotocol.io` | No |
-| `CLERK_SECRET_KEY` | `sk_live_j5y6TVVydxWe93qEj9DAUhB9jZJtpM8i4Ti4mVi5XQ` | No |
-| `STRIPE_SECRET_KEY` | from `.env` | No |
-| `STRIPE_WEBHOOK_SECRET` | `whsec_ASew8AShDg6NtD4jn4F6O1lY3FYv4SDz` | No |
+| `CLERK_SECRET_KEY` | `sk_live_...` (see Clerk dashboard) | No |
+| `STRIPE_SECRET_KEY` | `sk_live_...` (see Stripe dashboard) | No |
+| `STRIPE_WEBHOOK_SECRET` | `whsec_...` (see Stripe webhook config) | No |
 | `STRIPE_SCRIBE_PRO_PRICE_ID` | `price_1TMSB5GRBK524I7zhmUgLmfm` | No |
 | `STRIPE_SCRIBE_ENTERPRISE_PRICE_ID` | `price_1TMSB5GRBK524I7zCIU71YN6` | No |
-| `CRP_SCRIBE_JWT_SECRET` | `81ab043b53aece66d5347efe2de858b8f65817f8a740ac9ec386e86c058f0000` | No |
+| `CRP_SCRIBE_JWT_SECRET` | `<generate-your-own-random-secret>` | No |
 | `CRP_SCRIBE_DATA_DIR` | `/app/data` | No |
 | `CRP_SCRIBE_FRONTEND_DIR` | `/app/frontend/dist` | No |
 | `CRP_SCRIBE_BASE_URL` | `https://scribe.crprotocol.io` | No |
@@ -65,13 +70,13 @@ PYTHONUNBUFFERED=1
 
 ```
 CLERK_ISSUER=https://clerk.comply.crprotocol.io
-CLERK_SECRET_KEY=sk_live_M96E3pt7KwY3WnsIERdW7VHUbZ1NfLLnExvCyq3Qom
-STRIPE_SECRET_KEY=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
-STRIPE_WEBHOOK_SECRET=whsec_TzhIRl0XKlqfRVGlVUOI6MwZmzlLxhum
+CLERK_SECRET_KEY=sk_live_...
+STRIPE_SECRET_KEY=sk_live_...
+STRIPE_WEBHOOK_SECRET=whsec_...
 STRIPE_COMPLY_PRO_PRICE_ID=price_1TMSGAGRBK524I7zvSYEwg9S
 STRIPE_COMPLY_ENTERPRISE_PRICE_ID=price_1TMSGAGRBK524I7z9ebppShK
 STRIPE_COMPLY_CLOUD_PRICE_ID=price_1TMSGAGRBK524I7zaSuQMMzb
-CRP_COMPLY_JWT_SECRET=253bdb0c901c4cd4894ca4176c9195b66a914927c193adc09b7c7dfef2797cb5
+CRP_COMPLY_JWT_SECRET=<generate-your-own-random-secret>
 CRP_COMPLY_DATA_DIR=/app/data
 CRP_COMPLY_FRONTEND_DIR=/app/frontend/dist
 CRP_COMPLY_BASE_URL=https://comply.crprotocol.io
@@ -79,7 +84,10 @@ CRP_COMPLY_CORS_ORIGINS=https://comply.crprotocol.io
 PYTHONUNBUFFERED=1
 ```
 
-> `STRIPE_SECRET_KEY` — open `C:\Users\User\Desktop\crp-comply\.env` and copy the `sk_live_51TLbVk...` value.
+> **NEVER commit real secret values to this file.** Pull the actual values from
+> your local `.env` / secrets manager and paste them directly into Railway's
+> Variables UI -- never into a git-tracked file. This file documents variable
+> *names* and *shape* only.
 
 ### All Comply Variables at a Glance
 
@@ -87,13 +95,13 @@ PYTHONUNBUFFERED=1
 |---|---|---|
 | `VITE_CLERK_PUBLISHABLE_KEY` | `pk_live_Y2xlcmsuY29tcGx5LmNycHJvdG9jb2wuaW8k` | **YES — toggle ON** |
 | `CLERK_ISSUER` | `https://clerk.comply.crprotocol.io` | No |
-| `CLERK_SECRET_KEY` | `sk_live_M96E3pt7KwY3WnsIERdW7VHUbZ1NfLLnExvCyq3Qom` | No |
-| `STRIPE_SECRET_KEY` | from `.env` | No |
-| `STRIPE_WEBHOOK_SECRET` | `whsec_TzhIRl0XKlqfRVGlVUOI6MwZmzlLxhum` | No |
+| `CLERK_SECRET_KEY` | `sk_live_...` (see Clerk dashboard) | No |
+| `STRIPE_SECRET_KEY` | `sk_live_...` (see Stripe dashboard) | No |
+| `STRIPE_WEBHOOK_SECRET` | `whsec_...` (see Stripe webhook config) | No |
 | `STRIPE_COMPLY_PRO_PRICE_ID` | `price_1TMSGAGRBK524I7zvSYEwg9S` | No |
 | `STRIPE_COMPLY_ENTERPRISE_PRICE_ID` | `price_1TMSGAGRBK524I7z9ebppShK` | No |
 | `STRIPE_COMPLY_CLOUD_PRICE_ID` | `price_1TMSGAGRBK524I7zaSuQMMzb` | No |
-| `CRP_COMPLY_JWT_SECRET` | `253bdb0c901c4cd4894ca4176c9195b66a914927c193adc09b7c7dfef2797cb5` | No |
+| `CRP_COMPLY_JWT_SECRET` | `<generate-your-own-random-secret>` | No |
 | `CRP_COMPLY_DATA_DIR` | `/app/data` | No |
 | `CRP_COMPLY_FRONTEND_DIR` | `/app/frontend/dist` | No |
 | `CRP_COMPLY_BASE_URL` | `https://comply.crprotocol.io` | No |

{crprotocol-5.0.0 → crprotocol-5.1.0}/README.md

@@ -16,13 +16,13 @@
 <p align="center">
   <a href="LICENSE.md"><img src="https://img.shields.io/badge/Spec-CC_BY--SA_4.0-blue.svg" alt="Spec: CC BY-SA 4.0"></a>
   <a href="LICENSE.md"><img src="https://img.shields.io/badge/Code-ELv2-orange.svg" alt="Code: Elastic License 2.0"></a>
-  <img src="https://img.shields.io/badge/Spec_Version-4.0.0-brightgreen.svg" alt="Spec Version: 4.0.0">
+  <img src="https://img.shields.io/badge/Spec_Version-5.1.0-brightgreen.svg" alt="Spec Version: 5.1.0">
   <img src="https://img.shields.io/badge/RFC_2119-Conformant-orange.svg" alt="RFC 2119">
   <img src="https://img.shields.io/badge/Language_Neutral-JSON_Schema-yellow.svg" alt="Language Neutral">
-  <img src="https://img.shields.io/badge/Status-v4.0.0_WIP-yellow.svg" alt="Status: v4.0.0 WIP">
+  <img src="https://img.shields.io/badge/Status-v5.1.0_Stable-brightgreen.svg" alt="Status: v5.1.0 Stable">
   <a href="https://github.com/AutoCyber-AI/context-relay-protocol/actions"><img src="https://img.shields.io/github/actions/workflow/status/AutoCyber-AI/context-relay-protocol/ci.yml?label=CI" alt="CI"></a>
   <img src="https://img.shields.io/badge/python-3.10%2B-blue.svg" alt="Python 3.10+">
-  <img src="https://img.shields.io/badge/tests-1%2C537-brightgreen.svg" alt="1,537 tests">
+  <img src="https://img.shields.io/badge/tests-2%2C796-brightgreen.svg" alt="2,796 tests">
 </p>
 
 <p align="center">
@@ -1063,7 +1063,7 @@ Full error taxonomy with all codes in [§6.10.4](specification/02_CORE_PROTOCOL.
 
 | Language | Status | Package | Repository |
 |----------|--------|---------|------------|
-| **Python** | ✅ v3.1.1 / v4.0.0 WIP | `pip install -e ".[dev]"` | This repository |
+| **Python** | ✅ v5.1.0 | `pip install crprotocol` | This repository |
 | **TypeScript** | 📋 Planned | `npm install @crp/sdk` | `crp-typescript` |
 | **Rust** | 📋 Planned | `cargo add crp` | `crp-rust` |
 
@@ -1197,11 +1197,13 @@ All configuration is optional. CRP works with zero configuration if you pass you
 ## Roadmap
 
 ### Phase 1: Open Specification ✅
-### Phase 4: v4.0 Implementation ← **We are here**
+### Phase 5: v5.1 Agentic Execution ← **We are here**
 - [x] Publish CRP v2.0 specification (9 documents, ~19,200 lines)
 - [x] JSON Schema definitions for all API types
-- [ ] Reference SDK: Python (`pip install crprotocol`)
-- [ ] Benchmark results: CRP on vs. off across tasks and models
+- [x] Reference SDK: Python (`pip install crprotocol`)
+- [x] Positioned agentic loop + Tool Capability Fabric (SPEC-049/050)
+- [x] Multi-turn CSO relay and continuation anti-repetition
+- [ ] Benchmark results published to BENCHMARKS.md
 - [ ] arXiv technical report with empirical evaluation
 
 ### Phase 2: Ecosystem
@@ -1325,7 +1327,7 @@ See [LICENSE.md](LICENSE.md) for the full license text.
 ---
 
 <p align="center">
-  <strong>Context Relay Protocol v4.0.0 WIP</strong><br>
+  <strong>Context Relay Protocol v5.1.0</strong><br>
   Zero configuration. Unbounded input. Unbounded output. Amplified reasoning.<br>
   Better context at every scale. Honest degradation. Quality-tiered.<br>
   Peak quality. Every window.

{crprotocol-5.0.0 → crprotocol-5.1.0}/crp/__init__.py

@@ -16,6 +16,23 @@ Usage::
 
 from __future__ import annotations
 
+import os
+import sys
+
+# ── Windows OpenMP duplicate-runtime crash mitigation ────────────────────
+# On Windows, numpy/scipy/scikit-learn/torch each bundle their own OpenMP
+# runtime DLL. When several of them coexist in one process (as they do once
+# the optional [nlp]/[full] extras — sentence-transformers, GLiNER, torch —
+# are installed alongside numpy/scipy/scikit-learn), loading more than one
+# OpenMP runtime can trigger a native access violation (observed as a fatal
+# crash inside `torch`'s C extension during model loading, e.g.
+# `GLiNER.from_pretrained(...)`). This must be set BEFORE numpy/scipy/torch
+# are first imported anywhere in the process, so it lives at the very top of
+# this package's __init__, ahead of every other CRP import.
+if sys.platform == "win32":
+    os.environ.setdefault("KMP_DUPLICATE_LIB_OK", "TRUE")
+    os.environ.setdefault("OMP_NUM_THREADS", "1")
+
 from crp._version import __version__
 from crp.config import CRPConfig
 from crp.core.config import ConfigurationResolver

{crprotocol-5.0.0 → crprotocol-5.1.0}/crp/_version.py

@@ -2,4 +2,4 @@
 # Licensed under Elastic License 2.0 — see LICENSE.md for details.
 """CRP — Context Relay Protocol SDK."""
 
-__version__ = "5.0.0"
+__version__ = "5.1.0"

{crprotocol-5.0.0 → crprotocol-5.1.0}/crp/extraction/stage3_gliner.py

@@ -129,6 +129,21 @@ class GLiNERExtractor:
                 return None
 
         try:
+            import sys
+
+            if sys.platform == "win32":
+                # Extra hardening beyond the process-level env vars set in
+                # crp/__init__.py: also cap torch's own intra-op thread pool.
+                # Prevents a Windows-only native access violation caused by
+                # multiple OpenMP runtimes (numpy/scipy/sklearn/torch) racing
+                # during model weight loading — see crp/__init__.py comment.
+                try:
+                    import torch  # type: ignore[import-untyped]
+
+                    torch.set_num_threads(1)
+                except Exception:
+                    pass
+
             from gliner import GLiNER  # type: ignore[import-untyped]
 
             self._model = GLiNER.from_pretrained("urchade/gliner_base")  # type: ignore[assignment]

{crprotocol-5.0.0 → crprotocol-5.1.0}/crp/scan/github_app.py

@@ -124,12 +124,20 @@ class GithubAppClient:
     # ------------------------------------------------------------------
 
     def list_repos(self, installation_id: str | int) -> list[dict[str, Any]]:
-        """List repositories accessible to this installation."""
+        """List repositories accessible to this installation.
+
+        GitHub's endpoint for this is ``GET /installation/repositories``
+        (singular ``installation``, no ID in the path) -- the installation
+        is implied entirely by the installation access token used for auth.
+        There is no ``/installations/{id}/repositories`` route; using it
+        returns a 404 regardless of a valid installation_id/token.
+        """
         token = self.installation_token(installation_id)
-        url = f"{_GITHUB_API_BASE}/installations/{installation_id}/repositories"
+        url = f"{_GITHUB_API_BASE}/installation/repositories"
         headers = {
-            "Authorization": f"token {token}",
+            "Authorization": f"Bearer {token}",
             "Accept": "application/vnd.github+json",
+            "X-GitHub-Api-Version": "2022-11-28",
         }
         repos: list[dict[str, Any]] = []
         while url:

{crprotocol-5.0.0 → crprotocol-5.1.0}/crp/scan/remediation.py

@@ -263,7 +263,14 @@ class RemediationEngine:
         # NEVER commit to base_branch directly — use the PR branch from the proposal
         pr_branch = proposal.pr_branch
 
-        github = GithubAppClient()
+        try:
+            github = GithubAppClient.from_env()
+        except RuntimeError:
+            return (
+                f"[GitHub App not configured — set GITHUB_APP_ID + "
+                f"GITHUB_APP_PRIVATE_KEY (or GITHUB_APP_PRIVATE_KEY_PATH). "
+                f"Proposed branch: {proposal.pr_branch}]"
+            )
         token = github.installation_token(installation_id)
         # Split owner/repo from the repo full name
         owner_name, repo_name = repo.split("/", 1) if "/" in repo else ("", repo)