FylloCodeBLOCK
FylloCode 是一个开源桌面应用,把 Claude Code、Codex 等本地 Coding Agent 接进项目工作流。派任务、审方案、追溯每一次改动,积累下来的规范和结论,换一个 Agent 也能继承。
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
FylloCode
让 Agent 越来越懂你的项目 开源桌面应用,把 Claude Code、Codex 等本地 Coding Agent 接进项目工作流 派任务、审方案、追溯每一次改动,积累下来的规范和结论,换一个 Agent 也能继承
English · 文档 · 下载
背景
每个 Agent 会话一结束,代码留下来了,决策却丢失了 ......
- 三天后不知道这行代码为什么改。 Agent 帮你动了 100+ 个文件,
git blame只告诉你谁提交的,不告诉你当时的决策背景。 - 两个月后没人知道方案的设计依据。 Agent 给出了一个架构方向,其他候选方案为什么被放弃,这些推理过程全消失在当时的聊天窗口里。
- 每个新会话都要从头建立上下文。 相同的问题,每个 Agent、每次对话都要重新解释一遍项目约束、历史决策和禁忌操作。
- 不同 Agent 给出的写法互相打架。 同一个项目里,换一个 Agent 或换一次会话,架构取向、命名习惯、测试方式就可能变一套,没有任何东西约束它们保持一致。
这几个问题的根源相同:Agent 缺少一个持久的、结构化的项目治理层。 FylloCode 就是这个治理层。
核心机制
FylloCode 在你已有的代码库和研发工具链之上工作,不替代 IDE、CI/CD 或项目管理系统。它在这些系统上面加一层, 专门解决"如何持续用好 Agent,并把工作产生的上下文留下来"的问题。
研发系统(GitHub / 云效 / Jira ...) ↑ 回写任务结果 ┌──────────────────────────────┐ │ FylloCode │ ← 治理层 │ fyllo-specs · fyllo-cortex │ │ fyllo-spawn │ └──────────────────────────────┘ ↓ 约束 & 注入上下文 Coding Agent(任意) ↓ 代码库
d078ec1ca3c5OBSERVED · 2026-10-08Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add FylloCode --env FYLLO_MCP_AUTH_TOKEN=${FYLLO_MCP_AUTH_TOKEN} -- npx -y [email protected]{
"mcpServers": {
"FylloCode": {
"command": "npx",
"args": [
"-y",
"[email protected]"
],
"env": {
"FYLLO_MCP_AUTH_TOKEN": "${FYLLO_MCP_AUTH_TOKEN}"
}
}
}
}Exposed tools (79)
75 read · 4 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
Architecture | read | x |
Balanced | read | Default profile |
Bom | read | bom |
GitHub | read | GitHub 仓库与 PR |
GitLab | read | GitLab 仓库与 MR |
IPC | read | channel rules |
Jira | read | 问题与项目跟踪 |
Linear | read | 简洁的问题跟踪 |
MainOnly | read | main project doc |
O3 | read | Fast |
OwnerOnly | read | owner rules |
Plan | read | Think first |
Review | read | Review current changes |
Sentry | read | 错误追踪与性能监控 |
Slack | read | 团队消息与通知 |
Sonnet | read | Fast and capable |
Stream | read | Stream intermediate output |
Testing | read | test conventions |
Tricky | read | uses Array<T> and </guidelines> breakout |
Vercel | read | 前端应用部署 |
WorktreeOnly | read | worktree doc |
a | read | Project A |
active-entry | read | Active |
active-newer | read | Active entry |
api-contract-reference | read | Reference entry |
apply-change | write | |
archive-change | read | |
available_agents | read | |
b | read | Project B |
both | read | Initialize |
cancel_session | read | |
check_session_status | read | |
compact | read | Compact context |
create-plan | write | |
create-proposal | write | |
describe_workflow_schema | read | |
description | read | Describe command with enough detail to verify the tooltip keeps a bounded readable size. |
empty | read | |
entry | read | Description |
entry-name | read | Entry description |
explore | read | |
first-entry | read | First |
foreign-owner | read | Evidence belongs to another Folder |
full | read | |
guidelines | read | |
hint | read | |
init | read | Initialize |
knowledge | read | |
lineage | read | |
list_workflows | read | |
markstream-vue-theme-subscription | read | Keep theme subscription stable |
message-markdown-theme-subscription | read | Read before changing markstream-vue theme subscription wiring |
package-anchor-reference | read | Read before changing the package anchor contract |
plan | read | Plan |
prompt_to_agent | read | |
propose_workflow | read | |
read_response | read | |
reference-entry | read | Reference entry |
renderer-theme-subscription | read | Read before changing renderer markdown theme subscriptions |
repository-evidence | read | Owner-qualified repository evidence |
review | read | Review code |
right-name | read | Mismatched filename |
second-entry | read | Second |
suspect-entry | read | Suspect entry |
trigger_workflow | write | |
unknown-entry | read | Unknown entry |
workspace | read | |
云效 | read | 统一覆盖 Projex、Codeup 与 Flow 的研发平台。 |
云效流水线 | read | Flow — 持续集成与交付 |
云效项目 | read | Projex — 任务与需求管理 |
可观测性 | read | 监控、日志与告警 |
源代码控制 | read | 管理仓库。 |
源码管理 | read | 代码的版本管理 |
通讯 | read | 通知与协作 |
部署 | read | 应用部署目标 |
钉钉 | read | 通知群消息与审批提醒 |
阿里云 | read | ECS / SAE / 函数计算部署 |
阿里云日志服务 | read | SLS — 日志查询与告警分析 |
项目管理 | read | 任务与需求的来源和归宿 |
Trust audit
BLOCKgrade F · trust 57/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | WARN |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | FAIL |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (10 observation(s))
- Network
- declared (5 observation(s))
- Shell
- declared (1 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (25)
- [x] 2.3 创建 `src/shared/fyllo-action/schemas.ts`,迁移 `src/shared/schemas/fyllo-action.ts` 中的 payload schemas,并新增 `registerActionInputSchema`、`transitionActionInputSchema`(`fail` 命令含可选 `error`,最大 1000
15:08:44.560 › [acp-mapper] ← sessionUpdate: tool_call_update {"content":[{"content":{"text":"Error: Error during web fetch for \"https://example.com\": Invalid regular expression: /^[_A-Za-zÀ-ÖØ-öø-˿15:08:44.560 › [acp-mapper] → {"type":"tool_call_update","toolCallId":"toolu_bdrk_016rt6xdFYgNU3RGLzXZvTdV","status":"failed","content":"Error: Error during web fetch for \"https://example.com\": Invaicon.icns
const url = new URL(incoming.url ?? "/", "http://127.0.0.1");
logger.info(`[bundled-mcp-host] proxy ready url=http://127.0.0.1:${address.port}`);: `http://127.0.0.1:${currentHost.proxyPort}/mcp/${managed.registration.name}`;text = data.decode("utf-8", "replace").lstrip("").strip()const secret = "sensitive-payload-marker";
const secret = "sensitive-legacy-payload";
.prettierignore
.openspec.yaml
.openspec.yaml
.openspec.yaml
.openspec.yaml
CLAUDE.md
references/CLAUDE.md
import icon from "../../../resources/app-icon.png?asset";
export { scanGuidelines } from "../../../mcp-servers/fyllo-cortex/src/utils/scan-guidelines";export type { GuidelineEntry } from "../../../mcp-servers/fyllo-cortex/src/types/guideline";} from "../../../mcp-servers/fyllo-cortex/src/utils/knowledge";
} from "../../../mcp-servers/fyllo-cortex/src/utils/knowledge";
http://127.0.0.1:<proxyPort>/mcp/fyllo-specs
http://127.0.0.1:<proxyPort>/mcp/fyllo-cortex
raw = base64.b64decode(payload, validate=False)
Gates applied: no_behavioural_pass.
d078ec1ca3c5full audit observations/trust-audit/mcp-server/fioooooooo__fyllocode.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | d078ec1ca3c5 | BLOCK | F | 57 | first audit |
Questions
What is the FylloCode MCP server?
FylloCode 是一个开源桌面应用,把 Claude Code、Codex 等本地 Coding Agent 接进项目工作流。派任务、审方案、追溯每一次改动,积累下来的规范和结论,换一个 Agent 也能继承。
What tools does FylloCode expose?
79 in total: 75 read-only, 4 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is FylloCode safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (57/100) and found 3 critical or high issues in the source. Each one is listed on this page with the file and line it is on.
What credentials does FylloCode need?
It reads FYLLO_MCP_AUTH_TOKEN from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does FylloCode run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as FylloCode at 0.16.0-beta.1.
How current is this page?
The grade is for one exact copy of the source (d078ec1ca3c5), read on 2026-10-08. The repository is watched and re-audited when it changes.