gdmcode 0.1.1__tar.gz → 0.1.3__tar.gz

gdmcode-0.1.3/.gitignore

@@ -0,0 +1,60 @@
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.pyd
+.Python
+*.egg-info/
+dist/
+build/
+.eggs/
+
+# Virtual environments
+.venv/
+venv/
+env/
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+.mutmut-cache/
+
+# IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+
+# gdm runtime state
+.gdm_remote.json
+.gdm/
+.context-memory/
+
+# Eval results (generated artifacts)
+evals/results/*.db
+evals/results/swebench/
+
+# claw-code (external tool, not part of this project)
+claw-code/
+
+# Scratch / probe files
+_chk.py
+_edit_cli.py
+_edit_cmd.py
+_lineend.py
+_patch2.py
+_pathcheck.py
+_probe.py
+_probe2.py
+_test2.py
+_v.py
+_write_html.py
+_write_tests.py
+_write_tests2.py
+
+# Secrets — never commit API keys
+.env
+**/.env
+*.env
+

gdmcode-0.1.3/CONTRIBUTING.md

@@ -0,0 +1,101 @@
+# Contributing to gdm
+
+Thank you for contributing! gdm is MIT-licensed and welcomes issues, bug fixes, new tools, and
+documentation improvements.
+
+---
+
+## 1. Development setup
+
+```bash
+git clone https://github.com/GedeonMatabaro/gdmcode
+cd gdmcode
+pip install -e ".[dev]"
+gdm health          # verify the install
+```
+
+Python 3.12+ required. A virtual environment (or `pipx`) is recommended.
+
+---
+
+## 2. Code style
+
+gdm uses **ruff** for linting and **mypy** for type checking.
+
+```bash
+make lint           # ruff check src tests && mypy src
+```
+
+Before opening a PR, ensure `make lint` passes with no errors. Formatting follows ruff defaults.
+Do not add `# noqa` suppressions without a comment explaining why.
+
+---
+
+## 3. Testing
+
+```bash
+make test           # python -m pytest
+```
+
+Run the full test suite before submitting. New features must ship with at least one test.
+Unit tests live in `tests/`; fixtures are in `tests/conftest.py`.
+
+To run a single file:
+```bash
+python -m pytest tests/test_agent_tools.py -v
+```
+
+---
+
+## 4. PR process
+
+**Branch naming:**
+- `feat/<short-slug>` — new feature
+- `fix/<short-slug>` — bug fix
+- `docs/<short-slug>` — documentation only
+
+**Commit format** (conventional commits):
+```
+feat: add impact analysis gate
+fix(config): handle missing TOML section gracefully
+docs: update architecture diagram
+```
+
+**Requirements before merge:**
+1. `make lint` passes.
+2. `make test` passes.
+3. PR description explains *what* and *why* (not just *how*).
+4. New public APIs include type annotations and docstrings.
+
+---
+
+## 5. Adding a new tool
+
+gdm tools live in `gdmcode/tools/`. Follow this checklist:
+
+1. **Implement** — create `gdmcode/tools/my_tool.py` with a function matching the tool call signature.
+2. **Register** — add the tool to the `REGISTRY` in `gdmcode/tools/__init__.py`.
+3. **Audit action** — define an audit action constant in `gdmcode/enterprise/audit.py`
+   (e.g. `MY_TOOL_CALL = "my_tool.call"`).
+4. **Permission** — add the tool to the appropriate permission tier in `gdmcode/permissions.py` if
+   it performs file writes, network calls, or shell execution.
+5. **Test** — write at least one test in `tests/test_my_tool.py` covering the happy path and a
+   permission-denied path.
+6. **Docs** — add the tool to the CLI reference in `docs/cli-reference.md` if it has a
+   corresponding subcommand.
+
+---
+
+## 6. Release process
+
+1. Update `_VERSION` in `gdmcode/cli.py`.
+2. Commit: `git commit -m "chore: bump version to X.Y.Z"`.
+3. Tag: `git tag vX.Y.Z`.
+4. Push the tag — CI will build the wheel and publish to PyPI automatically via GitHub Actions.
+5. Verify with `pip install gdmcode==X.Y.Z && gdm version`.
+
+---
+
+## Questions?
+
+Open an issue or start a discussion on GitHub. We aim to respond within two business days.

gdmcode-0.1.3/PKG-INFO

@@ -0,0 +1,244 @@
+Metadata-Version: 2.4
+Name: gdmcode
+Version: 0.1.3
+Summary: gdm: AI coding agent for professional developers
+Project-URL: Homepage, https://github.com/guidegdm/gdmcode
+Project-URL: Repository, https://github.com/guidegdm/gdmcode
+Project-URL: Bug Tracker, https://github.com/guidegdm/gdmcode/issues
+Project-URL: Documentation, https://github.com/guidegdm/gdmcode#readme
+Project-URL: Changelog, https://github.com/guidegdm/gdmcode/blob/main/CHANGELOG.md
+License: MIT
+Keywords: ai,coding-agent,developer-tools,llm
+Classifier: Development Status :: 3 - Alpha
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Software Development :: Code Generators
+Classifier: Topic :: Utilities
+Requires-Python: >=3.11
+Requires-Dist: httpx>=0.27.0
+Requires-Dist: keyring>=25.0.0
+Requires-Dist: markdownify>=0.12.0
+Requires-Dist: openai>=1.30.0
+Requires-Dist: prompt-toolkit>=3.0.43
+Requires-Dist: pydantic>=2.7.0
+Requires-Dist: requests>=2.32.0
+Requires-Dist: rich>=13.7.0
+Requires-Dist: tiktoken>=0.7.0
+Requires-Dist: tomli-w>=1.0.0
+Requires-Dist: tree-sitter-javascript>=0.23.0
+Requires-Dist: tree-sitter-python>=0.23.0
+Requires-Dist: tree-sitter-typescript>=0.23.0
+Requires-Dist: tree-sitter>=0.23.0
+Requires-Dist: typer>=0.12.0
+Requires-Dist: websockets>=12.0
+Provides-Extra: all
+Requires-Dist: faiss-cpu>=1.8.0; extra == 'all'
+Requires-Dist: itsdangerous>=2.2.0; extra == 'all'
+Requires-Dist: numpy>=1.26.0; extra == 'all'
+Requires-Dist: openpyxl>=3.1.0; extra == 'all'
+Requires-Dist: pdfplumber>=0.11.0; extra == 'all'
+Requires-Dist: playwright>=1.44.0; extra == 'all'
+Requires-Dist: python-docx>=1.1.0; extra == 'all'
+Requires-Dist: qrcode>=7.4.0; extra == 'all'
+Requires-Dist: sounddevice>=0.4.6; extra == 'all'
+Requires-Dist: starlette>=0.37.0; extra == 'all'
+Requires-Dist: uvicorn>=0.29.0; extra == 'all'
+Requires-Dist: webrtcvad>=2.0.10; extra == 'all'
+Requires-Dist: websockets>=12.0; extra == 'all'
+Requires-Dist: xlrd>=2.0.1; extra == 'all'
+Provides-Extra: browser
+Requires-Dist: playwright>=1.44.0; extra == 'browser'
+Requires-Dist: websockets>=12.0; extra == 'browser'
+Provides-Extra: dev
+Requires-Dist: pytest-asyncio>=0.23.0; extra == 'dev'
+Requires-Dist: pytest-cov>=5.0.0; extra == 'dev'
+Requires-Dist: pytest-mock>=3.14.0; extra == 'dev'
+Requires-Dist: pytest>=8.2.0; extra == 'dev'
+Provides-Extra: docs
+Requires-Dist: openpyxl>=3.1.0; extra == 'docs'
+Requires-Dist: pdfplumber>=0.11.0; extra == 'docs'
+Requires-Dist: python-docx>=1.1.0; extra == 'docs'
+Requires-Dist: xlrd>=2.0.1; extra == 'docs'
+Provides-Extra: evals
+Requires-Dist: bandit[toml]>=1.7.0; extra == 'evals'
+Requires-Dist: datasets>=2.20.0; extra == 'evals'
+Requires-Dist: docker>=7.1.0; extra == 'evals'
+Requires-Dist: jinja2>=3.1.0; extra == 'evals'
+Requires-Dist: mypy>=1.10.0; extra == 'evals'
+Requires-Dist: pandas>=2.2.0; extra == 'evals'
+Requires-Dist: pyyaml>=6.0.0; extra == 'evals'
+Requires-Dist: ruff>=0.4.0; extra == 'evals'
+Requires-Dist: swebench>=3.0.0; extra == 'evals'
+Provides-Extra: faiss
+Requires-Dist: faiss-cpu>=1.8.0; extra == 'faiss'
+Provides-Extra: mobile
+Provides-Extra: plugin-dev
+Requires-Dist: build>=1.2; extra == 'plugin-dev'
+Requires-Dist: twine>=5.0; extra == 'plugin-dev'
+Provides-Extra: remote
+Requires-Dist: itsdangerous>=2.2.0; extra == 'remote'
+Requires-Dist: qrcode>=7.4.0; extra == 'remote'
+Requires-Dist: starlette>=0.37.0; extra == 'remote'
+Requires-Dist: uvicorn>=0.29.0; extra == 'remote'
+Requires-Dist: websockets>=12.0; extra == 'remote'
+Provides-Extra: sentry
+Requires-Dist: fastapi>=0.111; extra == 'sentry'
+Requires-Dist: httpx>=0.27; extra == 'sentry'
+Requires-Dist: uvicorn[standard]>=0.29; extra == 'sentry'
+Provides-Extra: voice
+Requires-Dist: numpy>=1.26.0; extra == 'voice'
+Requires-Dist: sounddevice>=0.4.6; extra == 'voice'
+Requires-Dist: webrtcvad>=2.0.10; extra == 'voice'
+Description-Content-Type: text/markdown
+
+# gdm — Open-Source AI Coding Agent
+
+A terminal-native coding agent that reads your codebase, reasons over it, and edits files autonomously — with configurable autonomy, local model support, and a full audit trail.
+
+![Build status](https://img.shields.io/badge/build-passing-brightgreen) ![PyPI](https://img.shields.io/badge/pypi-gdm--code-blue) ![License: MIT](https://img.shields.io/badge/license-MIT-green)
+
+---
+
+## Quick start
+
+```bash
+pipx install gdmcode
+gdm version
+gdm login grok        # or: gemini | codex
+cd my-project && gdm "fix the login bug"
+```
+
+If you cannot reach a model provider directly, start `gdm` and run `/proxy`, then paste the relay token with `/proxy token`. The provider API key stays encrypted locally; the relay should not persist provider keys.
+
+See [docs/quick-start.md](docs/quick-start.md) for a step-by-step walkthrough.
+
+---
+
+## Why gdm?
+
+- **Local-model support** — run fully air-gapped with Ollama or vLLM; no data leaves your machine.
+- **Autonomy slider** — five levels from "ask everything" to "fully autonomous"; start low, build trust.
+- **Open source and auditable** — MIT licensed, hash-chained audit log, team policy enforcement.
+
+---
+
+## Feature status
+
+| Feature | Status | Task(s) |
+|---|---|---|
+| Multi-model routing (Grok / Gemini / Codex) | ✅ implemented | models-001/002/005 |
+| Local model support — Ollama / vLLM | ✅ implemented | models-003 |
+| Cost tracking and budget caps | ✅ implemented | runtime-005 |
+| Session checkpoint / restore | ✅ implemented | core-002/003 |
+| Audit log with hash chain | ✅ implemented | enterprise-002 |
+| Team configuration and policy | ✅ implemented | enterprise-001 |
+| Web search (debug / research) | ✅ implemented | tools-003 |
+| Code index + whole-codebase mode | ✅ implemented | memory-001/005 |
+| Git workflow automation | ✅ implemented | autonomous-003 |
+| Verification loop | ✅ implemented | autonomous-002 |
+| Self-healing debug loop | ✅ implemented | autonomous-004 |
+| Multi-file orchestration | ✅ implemented | autonomous-001 |
+| Impact analysis gate | ✅ implemented | autonomous-005 |
+| Autonomy slider — 5 levels | ✅ implemented | ide-004 |
+| VS Code extension | ✅ implemented | ide-001/002/003 |
+| Voice interface (PTT + streaming) | ✅ implemented | voice-001/002/003 |
+| Remote / LAN server + phone UI | ✅ implemented | remote-001/002/003 |
+| Chrome extension + bridge | ✅ implemented | extensions-002 |
+| MCP protocol server | ✅ implemented | integrations-003 |
+| GitHub Actions CI integration | ✅ implemented | integrations-001 |
+| Playwright browser automation | ✅ implemented | tools-002 |
+| Sub-agent parallel execution | ✅ implemented | tools-005 |
+| RBAC (role-based access control) | 📋 planned | enterprise-003 |
+| Usage analytics dashboard | 📋 planned | enterprise-004 |
+| Hermetic sandbox | ✅ implemented | moonshot-002 |
+| Patch risk scoring | ✅ implemented | moonshot-006 |
+| Impact graph smart test selection | ✅ implemented | moonshot-007 |
+| CI eval pipeline | ✅ implemented | evals-002 |
+| Continuous memory | ✅ implemented | memory-003 |
+| Proxy/relay mode (geo-restricted regions) | ✅ implemented | proxy-001 |
+| OIDC identity integration | 📋 planned | — |
+
+---
+
+## Installation
+
+**Recommended (isolated environment):**
+```bash
+pipx install gdmcode
+```
+
+**pip:**
+```bash
+pip install gdmcode
+```
+
+**Development install:**
+```bash
+git clone https://github.com/GedeonMatabaro/gdmcode
+cd gdmcode
+pip install -e ".[dev]"
+gdm health
+```
+
+**Note:** The PyPI package is `gdmcode`. The command is `gdm`.
+
+---
+
+## Configuration
+
+gdm uses a 7-layer precedence system (highest wins):
+
+```
+CLI flags > env vars (GDM_*) > team policy > team preferences
+         > project .gdm/config.toml > user ~/.config/gdm/config.toml > defaults
+```
+
+Quick setup:
+```bash
+gdm login grok        # or: gemini | codex | all
+```
+
+Secrets are stored in the OS keychain — never in TOML files.
+
+Interactive setup is also available inside `gdm` with `/login` and `/proxy`. First launch without credentials shows onboarding instead of starting model agents.
+
+Full configuration reference: [docs/configuration.md](docs/configuration.md)
+
+---
+
+## Enterprise
+
+gdm includes audit and governance features suitable for team deployment:
+
+- **Audit log** — every tool call is logged with a hash chain for tamper detection.
+- **Team policy** — `.gdm/team.toml` enforces autonomy caps, network allowlists, and git permissions.
+- **RBAC** — planned (enterprise-003).
+
+See [docs/security-hardening.md](docs/security-hardening.md) for deployment guidance.
+
+---
+
+## Architecture
+
+gdm runs a tool-calling agent loop: the model receives a task, selects tools from a registered
+registry, executes them (subject to injection checks, RBAC, and network policy), and writes the
+results back into context. Sessions are checkpointed to SQLite; memory is indexed for whole-codebase
+reasoning.
+
+Full architecture diagram: [docs/architecture.md](docs/architecture.md)
+
+---
+
+## Contributing
+
+See [CONTRIBUTING.md](CONTRIBUTING.md). Issues and PRs welcome.
+
+---
+
+## License
+
+MIT — Copyright GedeonMatabaro

gdmcode-0.1.3/README.md

@@ -0,0 +1,147 @@
+# gdm — Open-Source AI Coding Agent
+
+A terminal-native coding agent that reads your codebase, reasons over it, and edits files autonomously — with configurable autonomy, local model support, and a full audit trail.
+
+![Build status](https://img.shields.io/badge/build-passing-brightgreen) ![PyPI](https://img.shields.io/badge/pypi-gdm--code-blue) ![License: MIT](https://img.shields.io/badge/license-MIT-green)
+
+---
+
+## Quick start
+
+```bash
+pipx install gdmcode
+gdm version
+gdm login grok        # or: gemini | codex
+cd my-project && gdm "fix the login bug"
+```
+
+If you cannot reach a model provider directly, start `gdm` and run `/proxy`, then paste the relay token with `/proxy token`. The provider API key stays encrypted locally; the relay should not persist provider keys.
+
+See [docs/quick-start.md](docs/quick-start.md) for a step-by-step walkthrough.
+
+---
+
+## Why gdm?
+
+- **Local-model support** — run fully air-gapped with Ollama or vLLM; no data leaves your machine.
+- **Autonomy slider** — five levels from "ask everything" to "fully autonomous"; start low, build trust.
+- **Open source and auditable** — MIT licensed, hash-chained audit log, team policy enforcement.
+
+---
+
+## Feature status
+
+| Feature | Status | Task(s) |
+|---|---|---|
+| Multi-model routing (Grok / Gemini / Codex) | ✅ implemented | models-001/002/005 |
+| Local model support — Ollama / vLLM | ✅ implemented | models-003 |
+| Cost tracking and budget caps | ✅ implemented | runtime-005 |
+| Session checkpoint / restore | ✅ implemented | core-002/003 |
+| Audit log with hash chain | ✅ implemented | enterprise-002 |
+| Team configuration and policy | ✅ implemented | enterprise-001 |
+| Web search (debug / research) | ✅ implemented | tools-003 |
+| Code index + whole-codebase mode | ✅ implemented | memory-001/005 |
+| Git workflow automation | ✅ implemented | autonomous-003 |
+| Verification loop | ✅ implemented | autonomous-002 |
+| Self-healing debug loop | ✅ implemented | autonomous-004 |
+| Multi-file orchestration | ✅ implemented | autonomous-001 |
+| Impact analysis gate | ✅ implemented | autonomous-005 |
+| Autonomy slider — 5 levels | ✅ implemented | ide-004 |
+| VS Code extension | ✅ implemented | ide-001/002/003 |
+| Voice interface (PTT + streaming) | ✅ implemented | voice-001/002/003 |
+| Remote / LAN server + phone UI | ✅ implemented | remote-001/002/003 |
+| Chrome extension + bridge | ✅ implemented | extensions-002 |
+| MCP protocol server | ✅ implemented | integrations-003 |
+| GitHub Actions CI integration | ✅ implemented | integrations-001 |
+| Playwright browser automation | ✅ implemented | tools-002 |
+| Sub-agent parallel execution | ✅ implemented | tools-005 |
+| RBAC (role-based access control) | 📋 planned | enterprise-003 |
+| Usage analytics dashboard | 📋 planned | enterprise-004 |
+| Hermetic sandbox | ✅ implemented | moonshot-002 |
+| Patch risk scoring | ✅ implemented | moonshot-006 |
+| Impact graph smart test selection | ✅ implemented | moonshot-007 |
+| CI eval pipeline | ✅ implemented | evals-002 |
+| Continuous memory | ✅ implemented | memory-003 |
+| Proxy/relay mode (geo-restricted regions) | ✅ implemented | proxy-001 |
+| OIDC identity integration | 📋 planned | — |
+
+---
+
+## Installation
+
+**Recommended (isolated environment):**
+```bash
+pipx install gdmcode
+```
+
+**pip:**
+```bash
+pip install gdmcode
+```
+
+**Development install:**
+```bash
+git clone https://github.com/GedeonMatabaro/gdmcode
+cd gdmcode
+pip install -e ".[dev]"
+gdm health
+```
+
+**Note:** The PyPI package is `gdmcode`. The command is `gdm`.
+
+---
+
+## Configuration
+
+gdm uses a 7-layer precedence system (highest wins):
+
+```
+CLI flags > env vars (GDM_*) > team policy > team preferences
+         > project .gdm/config.toml > user ~/.config/gdm/config.toml > defaults
+```
+
+Quick setup:
+```bash
+gdm login grok        # or: gemini | codex | all
+```
+
+Secrets are stored in the OS keychain — never in TOML files.
+
+Interactive setup is also available inside `gdm` with `/login` and `/proxy`. First launch without credentials shows onboarding instead of starting model agents.
+
+Full configuration reference: [docs/configuration.md](docs/configuration.md)
+
+---
+
+## Enterprise
+
+gdm includes audit and governance features suitable for team deployment:
+
+- **Audit log** — every tool call is logged with a hash chain for tamper detection.
+- **Team policy** — `.gdm/team.toml` enforces autonomy caps, network allowlists, and git permissions.
+- **RBAC** — planned (enterprise-003).
+
+See [docs/security-hardening.md](docs/security-hardening.md) for deployment guidance.
+
+---
+
+## Architecture
+
+gdm runs a tool-calling agent loop: the model receives a task, selects tools from a registered
+registry, executes them (subject to injection checks, RBAC, and network policy), and writes the
+results back into context. Sessions are checkpointed to SQLite; memory is indexed for whole-codebase
+reasoning.
+
+Full architecture diagram: [docs/architecture.md](docs/architecture.md)
+
+---
+
+## Contributing
+
+See [CONTRIBUTING.md](CONTRIBUTING.md). Issues and PRs welcome.
+
+---
+
+## License
+
+MIT — Copyright GedeonMatabaro

gdmcode-0.1.3/config.toml

@@ -0,0 +1,55 @@
+# config.toml — gdm application settings template
+#
+# Copy to ~/.config/gdm/config.toml for user-wide settings, or to
+# .gdm/config.toml in your project directory for project-level overrides.
+#
+# Precedence (lowest → highest):
+#   defaults → user config → project config → team preferences
+#   → team policy → environment variables → CLI flags
+#
+# Team [policy] values override environment variables and CLI flags.
+# Secrets (api_key, token, password, credential) are stored in the OS keychain
+# via `gdm login` — they must NOT appear in any config file.
+#
+# All keys are documented with type, default value, and description.
+
+# ─────────────────────────────────────────────────────────────────────────────
+# [models] — AI model selection
+# ─────────────────────────────────────────────────────────────────────────────
+
+[models]
+primary  = "grok-4-0106"   # default: "grok-4-0106"  — model for all tasks
+fallback = "o3-mini"       # default: "o3-mini"       — fallback when primary is unavailable
+
+# ─────────────────────────────────────────────────────────────────────────────
+# [budget] — cost controls
+# ─────────────────────────────────────────────────────────────────────────────
+
+[budget]
+session_limit_usd = 10.0    # type: float, default: 10.0   — max cost per session; 0 = unlimited
+monthly_limit_usd = 200.0   # type: float, default: 200.0  — monthly cap across all sessions
+
+# ─────────────────────────────────────────────────────────────────────────────
+# [autonomy] — agent autonomy level
+# ─────────────────────────────────────────────────────────────────────────────
+
+[autonomy]
+level = 2   # type: int (0–5), default: 2
+            # 0 = ask everything  |  3 = balanced  |  5 = fully autonomous
+            # Start low and increase as trust builds.
+
+# ─────────────────────────────────────────────────────────────────────────────
+# [audit] — audit log retention
+# ─────────────────────────────────────────────────────────────────────────────
+
+[audit]
+retain_days = 90   # type: int (≥1), default: 90 — delete audit entries older than N days
+
+# ─────────────────────────────────────────────────────────────────────────────
+# [analytics] — usage analytics and reporting
+# ─────────────────────────────────────────────────────────────────────────────
+
+[analytics]
+consent_required = false   # type: bool, default: false — require per-user opt-in
+roi_estimation   = false   # type: bool, default: false — enable estimated time-saved reporting
+anonymise        = true    # type: bool, default: true  — hash actor IDs (recommended)

gdmcode-0.1.3/docs/agentic-runtime-audit.md

@@ -0,0 +1,57 @@
+# Agentic Runtime Audit
+
+## Decisions locked in
+
+- The proxy must not persist user provider API keys server-side.
+- Server-side proxy state may keep only short-lived, non-reversible token fingerprints for abuse controls such as rate limiting.
+- The CLI must start as a lightweight shell before login so `/login`, `/proxy`, `/help`, and local commands are usable without constructing model clients or agent loops.
+- Agent/runtime initialization should be lazy: build the model client, router, transcript, orchestrator, and permission context only when the first model-backed turn starts.
+- First-run UX should be explicit: users should see the two valid connection paths immediately, not discover them only after a failed prompt.
+
+## Fixed in this pass
+
+- `gdm` can load config without credentials for interactive startup.
+- The REPL now supports `/login [provider]`, reloads config after login, and defers agent construction until the first non-command prompt.
+- `/proxy on` can be used before the agent exists; the next model turn uses the proxy configuration.
+- One-shot prompts fail fast with a clear login/proxy message instead of constructing a broken model client.
+- Proxy-only credentials are valid: an enabled proxy token can be the only credential path.
+- Proxy rate limiting no longer stores raw bearer tokens in memory; it stores SHA-256 fingerprints.
+- Proxy request handling rejects malformed `Content-Length`, oversized chunked bodies, invalid JSON, and non-object JSON before upstream forwarding.
+- Streaming proxy responses now preserve upstream HTTP status codes.
+- First-run REPL now shows a guided onboarding panel when no model connection is configured.
+- A plain prompt without credentials now re-shows the setup panel instead of a dead-end error.
+- Proxy setup copy now encourages `/proxy token` hidden input instead of pasting secrets inline into shell history.
+
+## Intended user journey
+
+1. User runs `gdm`.
+2. If credentials or proxy mode already exist, gdm shows the header and keeps startup light: the agent initializes on the first real prompt.
+3. If no model connection exists, gdm starts the REPL anyway and shows:
+   - Direct provider path: `/login grok`, `/login gemini`, or `/login codex`.
+   - Proxy path: `/proxy token` to enter the provider key hidden and store it locally, then `/proxy on`.
+4. User can still run `/help`, `/status`, `/proxy status`, `/doctor`, or `/exit` without an LLM.
+5. After login/proxy setup, the next non-command prompt initializes the model client and agent loop.
+6. Non-interactive `gdm --prompt ...` remains strict: it exits early with setup instructions if no model route exists.
+
+## Remaining high-impact gaps
+
+- Consolidate runtime state away from legacy project-local memory paths into a single documented `.gdm`/appdata boundary.
+- Add end-to-end CLI tests for a complete no-credential REPL session: `/help`, `/proxy token`, `/proxy on`, `/login`, then first model turn with a mocked model backend.
+- Add proxy integration tests with mocked upstream streaming responses for 200, 401, 429, and disconnect cases.
+- Audit background daemon startup paths so background agents never run unless explicitly requested or scheduled by a user-approved policy.
+- Add startup timing tests or benchmarks for `gdm --version`, `gdm health`, and bare `gdm` to prevent future eager imports.
+
+## Release hardening pass
+
+- Build configuration now excludes generated caches, coverage files, local runtime state, result databases, scratch files, and internal task/brainstorm docs from release artifacts.
+- The wheel intentionally contains only the runtime `src` package. The sdist includes runtime code, proxy code, tests, curated docs, and project metadata.
+- Tracked generated artifacts are being removed from version control, including `__pycache__`, `.pyc`, `.coverage`, and `evals/results/budget.db`.
+- Version display now uses the package `__version__` so installed metadata and `gdm version` stay aligned.
+- Test fixtures no longer contain full secret-shaped literals, reducing false positives during release artifact scans.
+- Release docs now describe direct login, relay setup, local credential storage, and the rule that provider API keys should not be persisted server-side.
+
+## Current release posture
+
+- Good for local dogfooding and an internal/private alpha build.
+- Acceptable for a public alpha only if the current top-level installed import package name, `src`, is explicitly accepted as temporary.
+- Not ideal for a polished public beta/stable PyPI release until the package is renamed to `gdmcode` or `gdm` and imports are migrated.