Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 7 additions & 4 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -148,9 +148,12 @@ Evolver integrates with major agent runtimes through `setup-hooks`. Run it once
#### Codex caveats

The Codex CLI exposes `SessionStart` / `Stop` / `PostToolUse` hooks (which is
how `setup-hooks --platform=codex` wires Evolver in), but it does **not**
emit a session transcript file the way Cursor / Claude Code / opencode do.
That means `evolver --review` cannot read raw session logs on Codex.
how `setup-hooks --platform=codex` wires Evolver in). Codex writes session
transcripts to `~/.codex/sessions/*.jsonl`, but the Stop hook payload does
not include a `transcript_path` field. Since v1.87.0, Evolver auto-discovers
transcripts under `~/.codex/sessions/` (and `~/.claude/projects/` for Claude
Code) without requiring manual `EVOLVER_CURSOR_TRANSCRIPTS_DIR` export, so
`evolver --review` can read raw session logs on Codex.

`setup-hooks --platform=codex` is lifecycle integration only; it does not route
Codex model requests through Evolver Proxy. To route Codex model traffic, run
Expand All @@ -160,7 +163,7 @@ command-backed auth runs `evolver proxy-token` or the absolute `node index.js
proxy-token --settings ...` helper emitted by `scripts/internal-proxy-env.sh
--codex-config` from a source checkout.

Evolver compensates by reading, in order:
For additional context, Evolver reads, in order:

1. `MEMORY.md` / `USER.md` in the workspace root (if you maintain them);
2. the `<!-- evolver-evolution-memory -->` section that
Expand Down
22 changes: 22 additions & 0 deletions README.zh-CN.md
Original file line number Diff line number Diff line change
Expand Up @@ -99,6 +99,28 @@ evolver setup-hooks --platform=claude-code

通过 `~/.claude/` 向 Claude Code 的 hook 系统注册 Evolver。安装完成后重启 Claude Code CLI。

#### Codex

```bash
evolver setup-hooks --platform=codex
```

写入 `~/.codex/hooks.json`,将 hook 脚本安装到 `~/.codex/hooks/`,并在 `config.toml` 中启用 `codex_hooks` 特性。重启 Codex CLI 后生效。

##### Codex 注意事项

Codex CLI 暴露 `SessionStart` / `Stop` / `PostToolUse` 钩子(`setup-hooks --platform=codex` 通过它们接入 Evolver)。Codex 会将会话转写写入 `~/.codex/sessions/*.jsonl`,但 Stop 钩子的 payload 不包含 `transcript_path` 字段。自 v1.87.0 起,Evolver 会自动发现 `~/.codex/sessions/`(以及 Claude Code 的 `~/.claude/projects/`)下的转写文件,无需手动 export `EVOLVER_CURSOR_TRANSCRIPTS_DIR`,因此 `evolver --review` 可以在 Codex 上读取原始会话日志。

`setup-hooks --platform=codex` 仅做生命周期集成,不会将 Codex 的模型请求路由到 Evolver Proxy。如需路由 Codex 的模型流量,请运行 Evolver Proxy,并为 Codex 配置一个用户级 OpenAI Responses 兼容的自定义 provider,其 `base_url` 指向 proxy 的 `/v1` 端点,鉴权命令运行 `evolver proxy-token` 或源码检出中 `scripts/internal-proxy-env.sh --codex-config` 输出的绝对路径 `node index.js proxy-token --settings ...` 辅助脚本。

除会话转写外,Evolver 还按以下顺序读取:

1. 工作区根目录的 `MEMORY.md` / `USER.md`(需自行维护);
2. `setup-hooks --platform=codex` 注入到项目 `AGENTS.md` 的 `<!-- evolver-evolution-memory -->` 段落;
3. 本地 `memory_graph.jsonl` 的尾部(Evolver 自行写入的每轮结果日志)。

如果以上来源在前几轮都还没有内容,你会看到 `memory_missing` / `user_missing` / `session_logs_missing` 作为 advisory 信号出现。随着 `memory_graph.jsonl` 积累结果,这些信号会自行消失——无需手动配置。

#### OpenClaw

OpenClaw 会识别 Evolver 向 stdout 输出的 `sessions_spawn(...)` 协议,**无需安装 hooks**。将 Evolver 克隆到 OpenClaw workspace 中,在会话内运行即可:
Expand Down
Loading