deepseek-bot 0.2.0__tar.gz

deepseek_bot-0.2.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 vicfei
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

deepseek_bot-0.2.0/PKG-INFO

@@ -0,0 +1,251 @@
+Metadata-Version: 2.4
+Name: deepseek-bot
+Version: 0.2.0
+Summary: DeepSeek Bot: self-evolving AI agent runtime — RSI under tiered authority, reversible ledger, capability security, crash-contained plugins
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/vicfei/deepseek-bot
+Project-URL: Repository, https://github.com/vicfei/deepseek-bot
+Project-URL: Issues, https://github.com/vicfei/deepseek-bot/issues
+Keywords: ai-agent,personal-agent,self-evolving,ledger,wechat,deepseek
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: zstandard>=0.23
+Requires-Dist: pydantic>=2.5
+Requires-Dist: pyyaml>=6
+Requires-Dist: httpx>=0.27
+Requires-Dist: aiohttp>=3.10
+Requires-Dist: segno>=1.6
+Requires-Dist: cryptography>=42
+Requires-Dist: croniter>=2
+Provides-Extra: dev
+Requires-Dist: pytest>=8; extra == "dev"
+Requires-Dist: hypothesis>=6; extra == "dev"
+Requires-Dist: ruff>=0.6; extra == "dev"
+Requires-Dist: pyright>=1.1.380; extra == "dev"
+Requires-Dist: pyyaml>=6; extra == "dev"
+Dynamic: license-file
+
+# DeepSeek Bot
+
+[![CI](https://github.com/vicfei/deepseek-bot/actions/workflows/ci.yml/badge.svg)](https://github.com/vicfei/deepseek-bot/actions/workflows/ci.yml)
+
+<p align="center"><img src="assets/banner.svg" alt="DeepSeek Bot — self-evolving personal agent, auditable by construction" width="880"></p>
+
+**A self-evolving AI agent runtime with a reversible, crash-recoverable core.**
+
+DeepSeek Bot is an AI agent that answers on your chat channels, turns experience
+into reusable skills, and treats *every* change it makes to itself as an
+auditable, revertible ledger entry — so autonomy never costs you control.
+
+[中文版 README](README.zh-CN.md)
+
+## Why it is different
+
+Most agent frameworks optimize for the demo. DeepSeek Bot optimizes for the other
+11 months: what happens when the agent modifies its own state, a plugin
+crashes at 3 a.m., a tool misbehaves, or you simply want yesterday back.
+The architecture is built around that, from the kernel up:
+
+- **Temporal composability — every change is reversible.** Any state change
+  registers its inverse; the system can roll back to any historical point in
+  LIFO order, including rolling back a rollback. Effects come in three
+  auditable forms (register / perform / bracket), all journaled.
+- **Spatial composability — declare, never wire.** Components declare
+  dependencies and coeffects reactively; the runtime drives lifecycles.
+  Shadow branches plug in without polluting the live system.
+- **Confluence as a tested invariant.** Any sequence of plugin
+  activation/deactivation converges to the same quiescent state as a
+  one-shot dependency-ordered composition. This is not a hope — it is
+  property-based tested, shrunk-counterexample-hardened.
+- **RSI with a leash — recursive self-improvement under tiered authority.**
+  Experience sediments into versioned skill assets (a Git-like content-addressed
+  tree); promotion to global scope requires shadow verification and
+  R-level authorization. Exploration happens in branches; merging happens
+  at gates. (Meta-improvement layer lands in M4.)
+- **A ledger, not a log.** An append-only, hash-chained ledger with
+  Merkle-sealed segments, snapshots, and a SQLite projection that can be
+  rebuilt from the log at any time. Crash recovery is fuzz-tested at
+  1,000 iterations — torn writes, mid-append deaths, and tampering all
+  recover or refuse loudly.
+- **Capability security end to end.** Peers derive trust levels
+  (owner / known / guest / untrusted); budgets, tool visibility
+  (R0 read-only … R4 destructive), and event visibility are enforced from
+  a single declarative table — a guest can use the agent without ever
+  seeing a write-capability or another session's events.
+- **Crash containment by construction.** External-source plugins execute in
+  isolated worker subprocesses: a segfault, hang, or memory leak cannot take
+  down the host; the call degrades, the worker is rebuilt, the task
+  continues.
+- **Real channels, real media.** Ships with a WeChat personal-bot channel
+  (text, typing indicator, chunked replies, images with on-device decryption
+  + CAS archiving + vision recognition, voice transcription) and a WebChat
+  console — all routed through one session-serializing gateway.
+
+## Architecture
+
+```
+┌──────────────────────────────────────────────────────────┐
+│ L3  Meta-improvement (RSI spark)                          │
+│     reflector · optimizer · R-level gates · shadow verify │
+├──────────────────────────────────────────────────────────┤
+│ L2  Experience assets                                     │
+│     skill library (version tree) · memory · curation      │
+├──────────────────────────────────────────────────────────┤
+│ L1  Execution                                             │
+│     agent loop · trace recorder · parallel sessions       │
+├──────────────────────────────────────────────────────────┤
+│ L0.5 Access plane                                         │
+│     channels · session routing · lane queues · trust      │
+├──────────────────────────────────────────────────────────┤
+│ L0  Composable kernel                                     │
+│     Context · reversible effects · coeffects · ledger     │
+└──────────────────────────────────────────────────────────┘
+```
+
+Key invariants enforced by the kernel (each backed by tests):
+
+| # | Invariant |
+|---|---|
+| INV-1 | Confluence — arbitrary activation orders converge |
+| INV-2 | Reversibility — rollback to any point, including rollbacks |
+| INV-3 | Monotonic capability decay down the spawn chain |
+| INV-4 | Single writer — shared state only via the serial ledger |
+| INV-5 | Constitution unreachable at runtime |
+| INV-6 | Mode ratchet — privileges auto-lower, never auto-raise |
+
+## The bargain
+
+<p align="center"><img src="assets/bargain.svg" alt="Autonomy is exchanged, never given: every capability buys its matching control" width="880"></p>
+
+## Lineage: DeepSeek Harness
+
+The architectural DNA of this project — **spatiotemporal composability**
+(reversible effects as the time axis, reactive coeffects as the space axis),
+**everything-is-a-plugin**, every plugin inspectable and replaceable — comes
+from the [Cordis](https://github.com/cordiverse/cordis) programming paradigm
+([_A Programming Paradigm for Spatiotemporal Composability_](https://arxiv.org/abs/2608.25512)),
+as pioneered in production by [**DeepSeek Harness** (`dsh`)](https://github.com/deepseek-ai/deepseek-harness),
+the open-source agent harness by DeepSeek AI.
+
+`DeepSeek Bot` is an **independent Python-route descendant** of that
+philosophy, applied to a self-evolving personal agent: no shared code with dsh
+(TypeScript), not a fork, and not affiliated with DeepSeek or the
+`deepseek-ai` organization. Where dsh is a general agent harness, DeepSeek Bot
+bets the same composability discipline on one scenario — a personal agent
+that runs for months, remembers with crypto-forgetting, schedules, promotes
+its own skills, and anchors its safety rules in a tamper-evident ledger.
+
+The first-class model adapter is DeepSeek (the project grew up debugging
+reasoning-model discipline against `deepseek` APIs), and the name pays the
+debt: **a DeepSeek bot, built the dsh way.**
+
+## Status
+
+| Milestone | Scope | Exit evidence |
+|---|---|---|
+| **M0 — composable kernel** ✅ | Context, effect triad, reactive coeffects, durable ledger, CAS, rollback | Confluence PBT 5/5; crash-recovery fuzz (1,000 iterations, zero inconsistencies); pyright strict; CI green |
+| **M1 — it runs** ✅ | LLM adapters, agent loop, skill sedimentation/reuse, WebSocket gateway + WebChat, traces & replay bundles | End-to-end dialogue, skill reuse, non-empty replay bundles — all with real LLM traffic |
+| **M2 — it connects** ✅ | WeChat channel (text/typing/media), trust-level degradation, worker isolation, hot plug/unplug | Real WeChat traffic (incl. image recognition & voice); trust matrix tested cell-by-cell; worker-crash penetration tests; bind→unbind→rebind without restart |
+| **M3 — it forks** ✅ | Sub-agent spawner (authorization algebra: strictly-decreasing budgets, tool subsets, trust ratchet), COW session branches, write-capture overlay | Real deepseek parent task spawning two parallel sub-agents (interleaved traces in the ledger); algebra pinned by PBT; 100-session write-capture SLO within bound |
+| **M4 — it evolves** ✅ | Promotion pipeline + shadow verification + R-gate, Curator, GEPA weekly loop, replay fidelity, model auto-routing, workflow compilation | Two real promoted skills on `main` (real shadow runs, −69% tokens on one); a full real GEPA cycle (quota failure → attribution → shadow tie → R1 promote); real routing downgrade (3 shadow wins → cluster downshift to the budget model, live savings dashboard); 50 real stable tasks → 9-node DAG compiled by deepseek → real execution with F.4 failure-degrade |
+| **M5 — hands off** 🔄 | Constitution layer (ledger-anchored, tamper ⇒ graceful shutdown), 30-day soak, security self-audit | Constitution anchored (sha `6b6291d…`) and re-verified daily; soak clock running since 2026-10-15 → 2026-11-14, **P0 = 0**; self-audit published in `docs/research/` |
+| **Memory system (§6.5)** ✅ | Episodic + semantic layers, trust-gated writes, poison lint, per-item DEK / per-subject KEK crypto-forgetting, cross-session recall injection | Real traffic: a real task distilled into encrypted memories; next task recalled them (`memory.use` in the ledger); owner CLI decrypts; ledger & CAS blobs verified plaintext-free |
+| **Scheduled jobs (Appendix C)** ✅ | Natural-language cron creation (one LLM compile), scheduler runtime, trust-snapshot revalidation (downgrade auto-pauses), misfire semantics, per-job rCeiling tool cap | Real traffic: “每天早上七点给我科技新闻摘要” compiled to `0 7 * * *` in one call; an every-minute job fired on schedule with real traces; daemon owns the ledger (CLI goes over loopback WS) |
+
+The authoritative design baseline lives in [`docs/design/`](docs/design/)
+(frozen, R3-graded; changes go through RFC). Research notes are in
+[`docs/research/`](docs/research/).
+
+## Quick start
+
+Requires Python 3.12+.
+
+```bash
+git clone https://github.com/vicfei/deepseek-bot.git
+cd deepseek-bot
+python3 -m venv .venv && .venv/bin/pip install -e ".[dev]"
+
+.venv/bin/pytest -q                    # full suite (unit + PBT + fuzz + channels)
+.venv/bin/ruff check src tests         # lint
+.venv/bin/pyright src                  # strict type check
+```
+
+Or install straight from git — the project is pip-packaged:
+
+```bash
+pip install git+https://github.com/vicfei/deepseek-bot.git
+```
+
+Running the daemon (gateway on loopback + WeChat channel) additionally needs a
+credentials file at `~/.deepseek-bot/credentials/providers.json` (0600) — see
+`src/deepseek_bot/llm.py` for the expected provider entries. (Existing
+`~/.naga` homes keep working: the runtime falls back to them automatically.)
+
+```bash
+deepseek-bot daemon start      # http://127.0.0.1:18789 — WebChat console
+deepseek-bot doctor            # self-check
+```
+
+## Try it
+
+Once the daemon is up:
+
+1. **WebChat** — open <http://127.0.0.1:18789> and chat directly (loopback
+   connection = owner trust).
+2. **WeChat** — click **📱 绑定微信** on the same page, scan the QR code with
+   WeChat, and the bot appears in your contact list — text, voice (auto-
+   transcribed) and images work end-to-end. Binding is hot: no restart needed.
+
+**Multiple people?** Yes — each person scans their own QR code and gets an
+**independent bot identity** (the platform provisions one per binding), with
+session, memory and trust isolated per user. Two concurrent real users have
+been verified in production. Note the current posture is *personal /
+small-circle deployment*: a fresh QR binding is granted **owner** trust on
+your instance, and the WebChat console is loopback-only by design — see
+`docs/design/` for the trust model before opening an instance to strangers.
+
+![demo: deepseek-bot doctor on the production daemon — ledger chain verified, constitution anchor matched](assets/demo-doctor.gif)
+
+*(Every line is real output from the live production daemon — running
+unattended since 2026-10-15.)*
+
+## Architecture
+
+<p align="center"><img src="assets/architecture.svg" alt="Channels → gateway lanes → agent loop with trust-gated tools → hash-chained ledger feeding CAS, projections and the constitution; evolution loops on the side" width="880"></p>
+
+## Repository layout
+
+```
+src/deepseek_bot/
+  kernel.py        # L0 seed kernel (frozen baseline twin; in-place evolution follows RFCs)
+  ledger_store.py  # durable ledger: append log, Merkle seals, snapshots, CAS, SQLite projection
+  gateway.py       # WebSocket gateway: lane queues, trust policy, event visibility matrix
+  agent.py         # agent loop with tracing and skill sedimentation
+  skills.py        # version-tree skill store (content-addressed)
+  trust.py         # peer trust derivation (allowlist + channel defaults)
+  worker.py        # subprocess isolation for external-source plugins
+  wechat.py        # WeChat iLink channel: polling, typing, chunking, media, vision
+  daemon.py cli.py # runtime host and CLI
+docs/design/       # frozen design baseline (4 documents + spec-constants + checker)
+docs/research/     # research notes and RFC material
+tests/             # unit, property-based, crash-fuzz, channel and integration suites
+```
+
+## Contributing
+
+Issues and pull requests are written in **English** (the project's working
+language for public artifacts; either language is fine in discussion).
+Changes go through issue → branch → PR → CI (ruff / pyright strict /
+pytest / freeze checker) → merge commit. The frozen design baseline under
+`docs/design/` is never edited in place — propose changes via RFC
+(see `docs/research/rfc-m2-registry-addendum.md` for the format).
+
+## License
+
+[MIT](LICENSE) © 2026 vicfei

deepseek_bot-0.2.0/README.md

@@ -0,0 +1,218 @@
+# DeepSeek Bot
+
+[![CI](https://github.com/vicfei/deepseek-bot/actions/workflows/ci.yml/badge.svg)](https://github.com/vicfei/deepseek-bot/actions/workflows/ci.yml)
+
+<p align="center"><img src="assets/banner.svg" alt="DeepSeek Bot — self-evolving personal agent, auditable by construction" width="880"></p>
+
+**A self-evolving AI agent runtime with a reversible, crash-recoverable core.**
+
+DeepSeek Bot is an AI agent that answers on your chat channels, turns experience
+into reusable skills, and treats *every* change it makes to itself as an
+auditable, revertible ledger entry — so autonomy never costs you control.
+
+[中文版 README](README.zh-CN.md)
+
+## Why it is different
+
+Most agent frameworks optimize for the demo. DeepSeek Bot optimizes for the other
+11 months: what happens when the agent modifies its own state, a plugin
+crashes at 3 a.m., a tool misbehaves, or you simply want yesterday back.
+The architecture is built around that, from the kernel up:
+
+- **Temporal composability — every change is reversible.** Any state change
+  registers its inverse; the system can roll back to any historical point in
+  LIFO order, including rolling back a rollback. Effects come in three
+  auditable forms (register / perform / bracket), all journaled.
+- **Spatial composability — declare, never wire.** Components declare
+  dependencies and coeffects reactively; the runtime drives lifecycles.
+  Shadow branches plug in without polluting the live system.
+- **Confluence as a tested invariant.** Any sequence of plugin
+  activation/deactivation converges to the same quiescent state as a
+  one-shot dependency-ordered composition. This is not a hope — it is
+  property-based tested, shrunk-counterexample-hardened.
+- **RSI with a leash — recursive self-improvement under tiered authority.**
+  Experience sediments into versioned skill assets (a Git-like content-addressed
+  tree); promotion to global scope requires shadow verification and
+  R-level authorization. Exploration happens in branches; merging happens
+  at gates. (Meta-improvement layer lands in M4.)
+- **A ledger, not a log.** An append-only, hash-chained ledger with
+  Merkle-sealed segments, snapshots, and a SQLite projection that can be
+  rebuilt from the log at any time. Crash recovery is fuzz-tested at
+  1,000 iterations — torn writes, mid-append deaths, and tampering all
+  recover or refuse loudly.
+- **Capability security end to end.** Peers derive trust levels
+  (owner / known / guest / untrusted); budgets, tool visibility
+  (R0 read-only … R4 destructive), and event visibility are enforced from
+  a single declarative table — a guest can use the agent without ever
+  seeing a write-capability or another session's events.
+- **Crash containment by construction.** External-source plugins execute in
+  isolated worker subprocesses: a segfault, hang, or memory leak cannot take
+  down the host; the call degrades, the worker is rebuilt, the task
+  continues.
+- **Real channels, real media.** Ships with a WeChat personal-bot channel
+  (text, typing indicator, chunked replies, images with on-device decryption
+  + CAS archiving + vision recognition, voice transcription) and a WebChat
+  console — all routed through one session-serializing gateway.
+
+## Architecture
+
+```
+┌──────────────────────────────────────────────────────────┐
+│ L3  Meta-improvement (RSI spark)                          │
+│     reflector · optimizer · R-level gates · shadow verify │
+├──────────────────────────────────────────────────────────┤
+│ L2  Experience assets                                     │
+│     skill library (version tree) · memory · curation      │
+├──────────────────────────────────────────────────────────┤
+│ L1  Execution                                             │
+│     agent loop · trace recorder · parallel sessions       │
+├──────────────────────────────────────────────────────────┤
+│ L0.5 Access plane                                         │
+│     channels · session routing · lane queues · trust      │
+├──────────────────────────────────────────────────────────┤
+│ L0  Composable kernel                                     │
+│     Context · reversible effects · coeffects · ledger     │
+└──────────────────────────────────────────────────────────┘
+```
+
+Key invariants enforced by the kernel (each backed by tests):
+
+| # | Invariant |
+|---|---|
+| INV-1 | Confluence — arbitrary activation orders converge |
+| INV-2 | Reversibility — rollback to any point, including rollbacks |
+| INV-3 | Monotonic capability decay down the spawn chain |
+| INV-4 | Single writer — shared state only via the serial ledger |
+| INV-5 | Constitution unreachable at runtime |
+| INV-6 | Mode ratchet — privileges auto-lower, never auto-raise |
+
+## The bargain
+
+<p align="center"><img src="assets/bargain.svg" alt="Autonomy is exchanged, never given: every capability buys its matching control" width="880"></p>
+
+## Lineage: DeepSeek Harness
+
+The architectural DNA of this project — **spatiotemporal composability**
+(reversible effects as the time axis, reactive coeffects as the space axis),
+**everything-is-a-plugin**, every plugin inspectable and replaceable — comes
+from the [Cordis](https://github.com/cordiverse/cordis) programming paradigm
+([_A Programming Paradigm for Spatiotemporal Composability_](https://arxiv.org/abs/2608.25512)),
+as pioneered in production by [**DeepSeek Harness** (`dsh`)](https://github.com/deepseek-ai/deepseek-harness),
+the open-source agent harness by DeepSeek AI.
+
+`DeepSeek Bot` is an **independent Python-route descendant** of that
+philosophy, applied to a self-evolving personal agent: no shared code with dsh
+(TypeScript), not a fork, and not affiliated with DeepSeek or the
+`deepseek-ai` organization. Where dsh is a general agent harness, DeepSeek Bot
+bets the same composability discipline on one scenario — a personal agent
+that runs for months, remembers with crypto-forgetting, schedules, promotes
+its own skills, and anchors its safety rules in a tamper-evident ledger.
+
+The first-class model adapter is DeepSeek (the project grew up debugging
+reasoning-model discipline against `deepseek` APIs), and the name pays the
+debt: **a DeepSeek bot, built the dsh way.**
+
+## Status
+
+| Milestone | Scope | Exit evidence |
+|---|---|---|
+| **M0 — composable kernel** ✅ | Context, effect triad, reactive coeffects, durable ledger, CAS, rollback | Confluence PBT 5/5; crash-recovery fuzz (1,000 iterations, zero inconsistencies); pyright strict; CI green |
+| **M1 — it runs** ✅ | LLM adapters, agent loop, skill sedimentation/reuse, WebSocket gateway + WebChat, traces & replay bundles | End-to-end dialogue, skill reuse, non-empty replay bundles — all with real LLM traffic |
+| **M2 — it connects** ✅ | WeChat channel (text/typing/media), trust-level degradation, worker isolation, hot plug/unplug | Real WeChat traffic (incl. image recognition & voice); trust matrix tested cell-by-cell; worker-crash penetration tests; bind→unbind→rebind without restart |
+| **M3 — it forks** ✅ | Sub-agent spawner (authorization algebra: strictly-decreasing budgets, tool subsets, trust ratchet), COW session branches, write-capture overlay | Real deepseek parent task spawning two parallel sub-agents (interleaved traces in the ledger); algebra pinned by PBT; 100-session write-capture SLO within bound |
+| **M4 — it evolves** ✅ | Promotion pipeline + shadow verification + R-gate, Curator, GEPA weekly loop, replay fidelity, model auto-routing, workflow compilation | Two real promoted skills on `main` (real shadow runs, −69% tokens on one); a full real GEPA cycle (quota failure → attribution → shadow tie → R1 promote); real routing downgrade (3 shadow wins → cluster downshift to the budget model, live savings dashboard); 50 real stable tasks → 9-node DAG compiled by deepseek → real execution with F.4 failure-degrade |
+| **M5 — hands off** 🔄 | Constitution layer (ledger-anchored, tamper ⇒ graceful shutdown), 30-day soak, security self-audit | Constitution anchored (sha `6b6291d…`) and re-verified daily; soak clock running since 2026-10-15 → 2026-11-14, **P0 = 0**; self-audit published in `docs/research/` |
+| **Memory system (§6.5)** ✅ | Episodic + semantic layers, trust-gated writes, poison lint, per-item DEK / per-subject KEK crypto-forgetting, cross-session recall injection | Real traffic: a real task distilled into encrypted memories; next task recalled them (`memory.use` in the ledger); owner CLI decrypts; ledger & CAS blobs verified plaintext-free |
+| **Scheduled jobs (Appendix C)** ✅ | Natural-language cron creation (one LLM compile), scheduler runtime, trust-snapshot revalidation (downgrade auto-pauses), misfire semantics, per-job rCeiling tool cap | Real traffic: “每天早上七点给我科技新闻摘要” compiled to `0 7 * * *` in one call; an every-minute job fired on schedule with real traces; daemon owns the ledger (CLI goes over loopback WS) |
+
+The authoritative design baseline lives in [`docs/design/`](docs/design/)
+(frozen, R3-graded; changes go through RFC). Research notes are in
+[`docs/research/`](docs/research/).
+
+## Quick start
+
+Requires Python 3.12+.
+
+```bash
+git clone https://github.com/vicfei/deepseek-bot.git
+cd deepseek-bot
+python3 -m venv .venv && .venv/bin/pip install -e ".[dev]"
+
+.venv/bin/pytest -q                    # full suite (unit + PBT + fuzz + channels)
+.venv/bin/ruff check src tests         # lint
+.venv/bin/pyright src                  # strict type check
+```
+
+Or install straight from git — the project is pip-packaged:
+
+```bash
+pip install git+https://github.com/vicfei/deepseek-bot.git
+```
+
+Running the daemon (gateway on loopback + WeChat channel) additionally needs a
+credentials file at `~/.deepseek-bot/credentials/providers.json` (0600) — see
+`src/deepseek_bot/llm.py` for the expected provider entries. (Existing
+`~/.naga` homes keep working: the runtime falls back to them automatically.)
+
+```bash
+deepseek-bot daemon start      # http://127.0.0.1:18789 — WebChat console
+deepseek-bot doctor            # self-check
+```
+
+## Try it
+
+Once the daemon is up:
+
+1. **WebChat** — open <http://127.0.0.1:18789> and chat directly (loopback
+   connection = owner trust).
+2. **WeChat** — click **📱 绑定微信** on the same page, scan the QR code with
+   WeChat, and the bot appears in your contact list — text, voice (auto-
+   transcribed) and images work end-to-end. Binding is hot: no restart needed.
+
+**Multiple people?** Yes — each person scans their own QR code and gets an
+**independent bot identity** (the platform provisions one per binding), with
+session, memory and trust isolated per user. Two concurrent real users have
+been verified in production. Note the current posture is *personal /
+small-circle deployment*: a fresh QR binding is granted **owner** trust on
+your instance, and the WebChat console is loopback-only by design — see
+`docs/design/` for the trust model before opening an instance to strangers.
+
+![demo: deepseek-bot doctor on the production daemon — ledger chain verified, constitution anchor matched](assets/demo-doctor.gif)
+
+*(Every line is real output from the live production daemon — running
+unattended since 2026-10-15.)*
+
+## Architecture
+
+<p align="center"><img src="assets/architecture.svg" alt="Channels → gateway lanes → agent loop with trust-gated tools → hash-chained ledger feeding CAS, projections and the constitution; evolution loops on the side" width="880"></p>
+
+## Repository layout
+
+```
+src/deepseek_bot/
+  kernel.py        # L0 seed kernel (frozen baseline twin; in-place evolution follows RFCs)
+  ledger_store.py  # durable ledger: append log, Merkle seals, snapshots, CAS, SQLite projection
+  gateway.py       # WebSocket gateway: lane queues, trust policy, event visibility matrix
+  agent.py         # agent loop with tracing and skill sedimentation
+  skills.py        # version-tree skill store (content-addressed)
+  trust.py         # peer trust derivation (allowlist + channel defaults)
+  worker.py        # subprocess isolation for external-source plugins
+  wechat.py        # WeChat iLink channel: polling, typing, chunking, media, vision
+  daemon.py cli.py # runtime host and CLI
+docs/design/       # frozen design baseline (4 documents + spec-constants + checker)
+docs/research/     # research notes and RFC material
+tests/             # unit, property-based, crash-fuzz, channel and integration suites
+```
+
+## Contributing
+
+Issues and pull requests are written in **English** (the project's working
+language for public artifacts; either language is fine in discussion).
+Changes go through issue → branch → PR → CI (ruff / pyright strict /
+pytest / freeze checker) → merge commit. The frozen design baseline under
+`docs/design/` is never edited in place — propose changes via RFC
+(see `docs/research/rfc-m2-registry-addendum.md` for the format).
+
+## License
+
+[MIT](LICENSE) © 2026 vicfei

deepseek_bot-0.2.0/pyproject.toml

@@ -0,0 +1,71 @@
+[build-system]
+requires = ["setuptools>=77"]  # PEP 639 SPDX license 表达式
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "deepseek-bot"
+version = "0.2.0"
+description = "DeepSeek Bot: self-evolving AI agent runtime — RSI under tiered authority, reversible ledger, capability security, crash-contained plugins"
+readme = "README.md"
+license = "MIT"
+license-files = ["LICENSE"]
+requires-python = ">=3.12"  # type 语句(递归类型别名)依赖 3.12;本机与 CI 均为 3.12
+keywords = ["ai-agent", "personal-agent", "self-evolving", "ledger", "wechat", "deepseek"]
+classifiers = [
+  "Development Status :: 4 - Beta",
+  "Intended Audience :: Developers",
+  "Programming Language :: Python :: 3.12",
+  "Programming Language :: Python :: 3.13",
+  "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+dependencies = [
+  "zstandard>=0.23",  # 快照 tar.zst(数据模型 §3.2);>=0.23 含类型标注
+  "pydantic>=2.5",    # 协议边界模型(Naga 纪律③;磁盘行 LedgerLine)
+  "pyyaml>=6",        # SKILL.md frontmatter(主文档 §6.1)
+  "httpx>=0.27",      # LLM/搜索适配器异步客户端(M1 批次 2)
+  "aiohttp>=3.10",    # WS 网关 + WebChat 静态页(M1 批次 3)
+  "segno>=1.6",       # 微信绑定二维码本地生成(M2-W1,不走公网第三方)
+  "cryptography>=42", # 微信媒体 AES-128-ECB 解密(M2-W3)
+  "croniter>=2",      # 定时任务 cron 解析与下次触发推算(附录 C)
+]
+
+[project.urls]
+Homepage = "https://github.com/vicfei/deepseek-bot"
+Repository = "https://github.com/vicfei/deepseek-bot"
+Issues = "https://github.com/vicfei/deepseek-bot/issues"
+
+[project.optional-dependencies]
+dev = [
+  "pytest>=8",
+  "hypothesis>=6",
+  "ruff>=0.6",
+  "pyright>=1.1.380",
+  "pyyaml>=6",
+]
+
+[project.scripts]
+deepseek-bot = "deepseek_bot.cli:main"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+# WebChat 页面是包内数据资源:缺此配置 wheel 会丢掉它(pip 装完网关 404)
+[tool.setuptools.package-data]
+deepseek_bot = ["webchat/*.html"]
+
+[tool.ruff]
+line-length = 100
+src = ["src", "tests"]
+# 冻结种子文件,原样平移(docs/design 基线的孪生副本,运行正确性由 pytest T 套件与
+# 冻结检查器 R14 把关);lint 只管 src/ 下的演进代码。2026 版 ruff 默认规则集
+# 含 I/UP/SIM/BLE/ASYNC,对逐字副本属于噪音,故整体排除而非逐条豁免。
+extend-exclude = ["src/deepseek_bot/kernel.py"]
+
+[tool.pyright]
+include = ["src"]
+typeCheckingMode = "strict"
+# 种子内核以文件级注释降级为 basic(见 kernel.py 首行):保持被分析状态,
+# 其导出符号对导入方类型可知;重写属后续内核产品化批次,届时删除该注释。
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]

deepseek_bot-0.2.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

deepseek_bot-0.2.0/src/deepseek_bot/__init__.py

@@ -0,0 +1,3 @@
+"""DeepSeek Bot:自进化个人智能体运行时(前代号 naga,2026-11 定名)。"""
+
+__version__ = "0.2.0"