feat(config)!: one ordering for configuration sources, and a bootstrap deny rule

$DSH_HOME/.env had just become an ordinary environment layer, which left the
harness resolving user-facing values from a flattened process.env that could
no longer say where a value came from. A key stored through the web page
stayed shadowed by an older key in the user's own .env. An endpoint could be
redirected by the project: the invoking directory's .env is materialized like
every other layer, and a base URL decides where a resolved API key is sent, so
a DEEPSEEK_BASE_URL written into a model-editable workspace would send the
user's credential — and the prompts carrying their code — to whatever host
that file named.

Give every user-facing value one ordering, with four kinds of source:

  explicit for this run     per-operation override, CLI argument
  > authored by deployment  --config / --config-replace
  > this launch's shell     inherited process environment
  > product-managed store   settings.yaml, .credentials.yaml
  > discovered file         $DSH_HOME/.env
  > defaults                schema default, shipped base, public default

The domains differ only in which tiers exist. The earlier split — credentials
ranking the environment over the managed file while settings ranked over the
environment — was inconsistent: the distinguishing fact is who authored the
source, not the domain.

packages/util/environment owns an immutable snapshot with per-layer
provenance. getFrom(name, sources) searches only the layers a caller names,
and omitting one is a refusal rather than a demotion: the adapters ask for
['process', 'user-env'], so no reordering can let a project file back into a
decision it was excluded from.

isBootstrapOnly rejects, before anything is materialized, any .env setting a
variable that governs how a process launches (PATH, SHELL, NODE_OPTIONS,
LD_PRELOAD), where code or model-visible instructions load from (the whole
DSH_* namespace, HOME, XDG_*), or how the network is reached (proxy and CA
variables). The namespace is denied wholesale so a switch added later cannot
become settable by being forgotten, and there is no opt-out.

verify-config-source-ownership keeps both rules: no unregistered process.env
read under packages/*/*/src (26 allowlisted with reasons), and no apiKey,
baseURL, or headers inlined from the environment in shipped Cordis config —
removing those inlines is what makes the deployment tier meaningful.
This commit is contained in:
Yichen Jiang
2026-08-04 16:17:32 +08:00
parent 8ddc53f7a0
commit 0512b12714
59 changed files with 1241 additions and 165 deletions

View File

@@ -9,11 +9,13 @@
import { pathToFileURL } from 'node:url'
import { readFileSync } from 'node:fs'
import { basename, dirname, resolve } from 'node:path'
import { parse as parseDotenv } from 'dotenv'
import * as yaml from 'js-yaml'
import { Context, type FiberState } from 'cordis'
import Loader, { type Entry, type EntryOptions } from '@cordisjs/plugin-loader'
import Include, { applyEntryPatches, entryListSchema, type PatchOptions } from '@cordisjs/plugin-include'
import { dshHomePath, resolveDshHome } from '@deepseek-ai/dsh-paths'
import { createEnvironmentSnapshot, isBootstrapOnly, type EnvironmentSnapshot } from '@deepseek-ai/dsh-environment'
import type {} from '@cordisjs/plugin-hmr'
// Side-effect type import: resolves `ctx.get('systemPrompt')` to the service.
import type {} from '@deepseek-ai/dsh-system-prompt'
@@ -66,12 +68,57 @@ export function loadEnv(
}
/**
* Load the dsh product CLI's user environment: the invoking directory's `.env`
* Parse one directory's `.env` without applying it, rejecting any bootstrap
* variable it declares. A discovered file must not decide how this process
* launches, where its code and model-visible instructions come from, or how it
* reaches the network, so a violation fails the launch BEFORE anything is
* materialized — reporting it afterwards would leave the process already
* running under the value it refused.
* @param binName - the diagnostic prefix on the thrown error.
* @param dir - the directory whose `.env` to read.
* @param warn - sink for the one-line unreadable-file diagnostic.
* @returns the parsed entries, or `undefined` when the file is absent or unreadable.
* @throws when the file declares a name {@link isBootstrapOnly} rejects.
*/
function readEnvLayer(
binName: string, dir: string, warn: (line: string) => void,
): { path: string; values: Record<string, string> } | undefined {
const path = resolve(dir, '.env')
let content: string
try {
content = readFileSync(path, 'utf8')
} catch (error) {
if ((error as NodeJS.ErrnoException | null)?.code !== 'ENOENT') {
warn(`${binName}: failed to load .env: ${String(error)}\n`)
}
// ENOENT (no .env) is fine — rely on the ambient environment.
return undefined
}
const values = parseDotenv(content)
for (const name of Object.keys(values)) {
if (!isBootstrapOnly(name)) continue
throw new Error(
`${binName}: ${path} sets "${name}", which only the launching environment may set`
+ ' (it decides how this process starts, where its code and instructions load from, or how it'
+ ` reaches the network); export ${name} instead of putting it in a .env file`,
)
}
return { path, values }
}
/**
* Load the dsh product CLI's user environment and return it as a snapshot that
* remembers which layer supplied each value: the invoking directory's `.env`
* over the Harness home's `.env`, both under the inherited process
* environment. `process.loadEnvFile` never replaces a name that is already
* set, so loading the project file first and the user file second is what
* makes the layering `user < project < inherited`; the app-boot tests pin all
* three layers because that ordering is the whole contract.
* environment.
*
* Each layer is parsed and checked before anything is applied, then applied in
* the order that makes the layering `user < project < inherited` —
* `process.loadEnvFile` never replaces a name already set. Values do reach
* `process.env`, because a user's own `--config` tree and third-party
* libraries read it; the returned snapshot is the authority for everything the
* harness itself resolves, since `process.env` alone cannot say whether a
* value came from the launching shell or from a file inside the workspace.
*
* The Harness home is resolved from the inherited environment *before* either
* file loads, so a project `.env` can never redirect which user document is
@@ -82,17 +129,28 @@ export function loadEnv(
* These are ordinary environment values with ordinary environment reach. A
* secret the Harness should own and isolate belongs in the credentials
* document, which is never materialized here.
* @param binName - the diagnostic prefix on the warn lines.
* @param binName - the diagnostic prefix on the diagnostics.
* @param cwd - the invoking directory whose `.env` is the project layer.
* @param warn - sink for the one-line misconfiguration diagnostics.
* @returns this run's frozen environment snapshot.
* @throws when either file declares a bootstrap-only variable.
*/
export function loadLayeredEnv(
binName: string, cwd: string = process.cwd(),
warn: (line: string) => void = line => void process.stderr.write(line),
): void {
): EnvironmentSnapshot {
const home = resolveDshHome()
loadEnv(binName, cwd, warn)
loadEnv(binName, home, warn)
const inherited = { ...process.env } as Record<string, string>
// Parse both layers first: a rejection must not leave one file applied.
const project = readEnvLayer(binName, cwd, warn)
const user = home === resolve(cwd) ? undefined : readEnvLayer(binName, home, warn)
if (project !== undefined) process.loadEnvFile(project.path)
if (user !== undefined) process.loadEnvFile(user.path)
return createEnvironmentSnapshot([
{ source: 'process', values: inherited },
...project === undefined ? [] : [{ source: 'project-env' as const, path: project.path, values: project.values }],
...user === undefined ? [] : [{ source: 'user-env' as const, path: user.path, values: user.values }],
])
}
/**