Files
gstack/browse/src/token-registry.ts
Garry Tan df89475b17 v1.91.11.0 refactor: one state-root rule, browse route table, shared shard engine, PTY harness split, MECE review resolvers (#3002)
* refactor(resolvers): split review.ts into MECE resolver modules (pure move)

Move every function from scripts/resolvers/review.ts, unchanged, into:
- review-dashboard.ts: review dashboard, plan-file review report
- plan-gates.ts: approval check, exit-plan-mode gate, plan-file discovery,
  plan-completion audit/gate (ship + review), plan verification exec
- spec-review.ts: both spec review loops, benefits-from, anti-shortcut clause
- outside-voice-steps.ts: Codex second opinion, adversarial step, Codex plan
  review, Codex doc review, disabled-outside record
- review-scope.ts: scope drift, cross-review dedup, shared-code reuse

review.ts is deleted; index.ts imports the new modules. gen-skill-docs
output is byte-identical for every host (--host all). Test imports and
source-path references are re-pointed; the two source-text report/gate
tests in gen-skill-docs.test.ts become behavioral renders across every
consuming skill and host. All 46 touchfile entries that named review.ts
now name all five modules, guarded by a recorded selection golden.

* test(browse): black-box auth matrix for every server route and both surfaces

Drives buildFetchHandler fetchLocal/fetchTunnel with no token, wrong token,
root token, scoped token and the SSE cookie for all 33 routes, plus unmatched
paths and wrong methods. Denials assert today's exact status, body and content
type; allowed credentials assert the handler was reached. Written against the
unchanged if-chain server so the W3 route-table refactor must keep it green.

* refactor(shard-engine): move scripts/test-strict-output.ts to scripts/lib/shard-engine.ts

The shared shard engine grows from the existing strict-output module
(runShardChild, killProcessGroup, signal forwarding, strict classifier).
scripts/test-strict-output.ts stays as a re-export so existing importers,
mock.module paths and the strict-output/run-shard-child tests are unchanged.
The engine inherits the global touchfile entry; the free runner's CLI-routing
fixture copies the new module.

* refactor(resolvers): decompose the three >150-line review resolvers (output-neutral)

Split generateAdversarialStep, generateCodexPlanReview and
generatePlanCompletionAuditInner into per-section helpers whose template
literals are copied verbatim, so every function in the new modules is at
or under 150 lines. gen-skill-docs output is byte-identical for every host
(--host all, compared against 96764e80 with a fixed --link-root).

* refactor(resolvers): one outside-voice failure policy (deliberate prose unification)

outsideVoiceFailurePolicy(ctx, opts) in outside-voice.ts now renders the
auth / timeout / empty-response bullets for all four call sites that
hand-typed them (Codex second opinion, adversarial step, Codex plan
review, design outside voices). Options are explicit per site
(timeoutMinutes, onTimeout, stderrOnEmpty, fallback, escape) with no
defaults.

Deliberate generated-prose changes (every host):
- office-hours: 'Fall back to <native> subagent.' becomes
  'Fall back to the <native> subagent below.'
- plan-devex-review: the plain 'Auth failure (stderr contains ...)'
  bullets become the canonical bold bullets; auth also triggers on
  'API key'; 'auth failed' becomes 'authentication failed'.
- review/ship adversarial: 'exceeded 9 minutes and was terminated'
  becomes 'timed out after 9 minutes and was terminated'; the timeout
  is still MISSING COVERAGE.
- design outside voices: unchanged.

Adds ratchet (d) (test/outside-voice-failure-policy.test.ts) with a
reasoned allowlist for /codex's own CLI errors, the MISSING COVERAGE
retention test, refreshed codex/factory ship goldens, and outside-voice.ts
in every touchfile entry of review.ts and design.ts (selection golden
extended).

* test(pty): fake PTY session driver with an injectable clock through the runner launch seam

The three plan-skill runners take an optional PtyDriver (launch, now,
monotonic, sleep); omitted, they use the real launcher and clocks exactly as
before. test/helpers/pty/fake-session.ts feeds scripted frames through that
seam, and claude-pty-runner.runners.unit.test.ts runs observation, counting
and floor for success, deadline timeout, permission prompt and plan-ready
outcomes with no CLI or real timers. These cases must stay green unchanged
through the W4 split and the runPtySession extraction.

Touchfiles: every entry that lists claude-pty-runner.ts or pty-screen.ts now
also lists test/helpers/pty/**.

* refactor(shard-engine): run both lanes on the shared engine; lane policy injected

Engine (scripts/lib/shard-engine.ts) gains the W2 primitives: per-shard
tmp/Chromium sandbox + async cleanup backstop, log-path allocation and
full-stream log capture, one duration-seed reader/writer with a lane
predicate, LanePolicy (seed predicate + zero-execution verdict),
strictShardStatus, and the shared CLI flag loop. runShardChild takes an
optional companion (signal/settle) and waits a bounded 250ms to reap a
wall-killed child.

Free lane stops spawning shards itself: runFreeShard uses runShardChild
with trackShardBrowser as the companion (win32 path unchanged: no process
group, no negative-pid kill). Its sync state-dir removal stays lane policy.
Paid lane uses the sandbox, log, seed, verdict and flag primitives; the
hollow-shard guard applies PAID_LANE_POLICY. Lane outcomes are unchanged
(free keeps >= 0 seeds and file-count zero-exec rule; paid keeps > 0 seeds,
warning under selection and passed-empty under EVALS_ALL).

paid-free-boundary's closure assertion now names the engine module, where
the strict classifier lives.

* test(shard-engine): engine unit tests, fixture-corpus equivalence, per-lane CLI parity

- test/shard-engine.test.ts: failing/unhandled/module-load output fails both
  lanes, per-lane zero-execution and seed rules, whole-group kill on a wall
  timeout (both lanes), mocked-win32 path with no negative-pid kill,
  companion settle order, log capture, sandbox isolation, flag loop.
- test/shard-engine-equivalence.test.ts + test/fixtures/shard-equivalence:
  seven outcome fixtures plus one real shard, run through both lanes and
  compared with classifications recorded from the base runners (96764e80).
- test/shard-cli-parity.test.ts + test/fixtures/shard-cli-parity: flag set,
  defaults, validation errors and the Unknown argument error per lane match
  the base runners.

* refactor(shard-engine): decompose runFreeShard and runPaidShard to <= 150 lines

Output-neutral extraction under the fixture-corpus equivalence and runner
tests: captureFreeStream, explainFreeVerdict and logFreeRecovery (free);
paidShardCommand, settleShardSpool, settleBootstrapRetention and
printLogTail (paid). The bootstrap scope-creation block that
bootstrap-retention.test.ts evaluates stays verbatim.

* refactor(pty): split claude-pty-runner.ts into test/helpers/pty/* behind a barrel

Pure move: every line of the former 5,047-line runner lands verbatim in one
module (four private helpers gain `export` for cross-module use):
binary, screen (absorbs test/helpers/pty-screen.ts, which now re-exports it),
launch, session (PtyDriver), judge, classify, auq, plan-native, boundaries,
runners/{observation,counting,floor}. claude-pty-runner.ts re-exports the
original public surface by name; pty/ modules import siblings directly.

Tests that read the runner's source text:
- rewritten as behavioral: the unit test's model-pin tripwire (fake CLI argv:
  fallback chain, --model before extraArgs, hermetic --strict-mcp-config),
  pty-skill-seeding-wiring (runners through the fake driver; launcher through
  a fake CLI reporting CLAUDE_CONFIG_DIR). The "three wrappers forward model"
  grep is replaced by the runners' fake-driver launch assertions.
- pty-screen-session / pty-screen-supervision: stop copying runner source;
  they mock.module the real pty/screen.ts (and the fixture cleanup) instead.
- re-pointed to the owning module (they execute a sliced runner body with
  injected boundaries; no seam exists for those boundaries yet):
  eng-seeded-completion-ai, plan-floor-permission, plan-create-prepublication,
  plan-count-completion; hermetic-wiring's source guard now reads pty/launch.ts
  and scans every pty/ module for raw process.env spreads.
- plan-count-timeout and pty-output-wake mock the viewport at pty/screen.ts.

* test(ratchet-c): enforcing module/function size ratchet and moved-code touchfile coverage

Ratchet (c) ships enforcing: test/helpers/module-size.ts counts file and
top-level function lengths by brace matching over masked source (strings,
comments, regex literals and template text masked; ${} expressions kept),
covering function declarations, arrow functions assigned to consts and
route-table handler properties, with no parser dependency. Its self-test
uses template literals and code-fence braces copied from
scripts/resolvers/review.ts and design.ts. test/fixtures/module-size-ratchet.json
binds scripts/lib/shard-engine.ts (<= 800 lines, <= 150 per function) and
records the residual runner sizes (free 2352, paid 1921) as non-growth caps;
allowlist entries are keyed on file plus matched text and need a reason.
Failure output lists file:line, the rule, Fix: and the allowlist path.

touchfiles.test.ts gains the moved-code superset check over
test/fixtures/touchfile-move-goldens/ (W2 golden recorded at 96764e80:
test-strict-output.ts and test-paid-shards.ts global, test-free-shards.ts none).

* refactor(browse): declared route table replaces the buildFetchHandler if-chain

The ~1,300-line if-chain in buildFetchHandler becomes a route table:
each entry declares method, path, auth kind and surfaces, and one auth
gate in browse/src/routes/table.ts returns the per-kind denial (root-bearer,
scoped, root-or-sse-cookie: 401 Unauthorized; root-token: 403 Root token
required; extension-origin: 403 Forbidden). Unmatched requests take the
declared fallthrough (root-bearer check, then plain-text 404). Handlers move
to browse/src/routes/{core,pairing,pty,tokens,tunnel,activity,commands,files,
inspector}.ts and receive a RouteContext with auth checks as functions
instead of closing over factory locals. Dispatch order is unchanged:
tunnel filter, beforeRoute overlay, gate, handler. TUNNEL_PATHS stays a
literal in server.ts.

Behavior-preserving: the black-box auth matrix from the previous commit
passes unchanged. /memory and /inspector/events are declared root-bearer
because the blanket check always ran before their SSE-cookie branch.

Source-text route tests are rewritten as behavioral tests through
buildFetchHandler or a route's real handler with a stub RouteContext
(browse/test/route-test-harness.ts). Checks with no runtime seam are
re-pointed to the route modules: Surface type, /inspector/events SSE
helper, sanitizeReplacer imports, /pty-inject-scan sidecar-client import,
and the ngrok config lookup and startTunnel wiring that stay in server.ts.

* test(browse): stubbed-handler auth matrix and route inventory for the route table

Every ROUTES entry runs through the real dispatcher and gate with stub
handlers on each declared surface and six credentials; denials assert the
exact status and body each auth kind returned at 96764e8, admitted
credentials assert the handler ran (with the gate's TokenInfo for scoped
routes). Also pins the reviewed route inventory (method, path, auth kind,
surfaces), that every entry declares auth and surfaces, that the table's
tunnel paths equal the TUNNEL_PATHS literal with GET /connect admitted, the
unmatched fallthrough, and that the root token is rejected on every tunnel
route through buildFetchHandler.

* test(browse): ratchet (b) keeps route dispatch inside the route table

Scans browse/src/server.ts and browse/src/routes/*.ts for pathname
comparisons; only the table matcher and the tunnel-surface filter are
allowed, listed with reasons in browse/test/fixtures/route-dispatch-allowlist.json
(keyed on file plus line text). Also checks every entry declares auth and
surfaces and that gstack registers no beforeRoute overlay itself. Self-tests
plant a violation and assert the file:line, Fix: and allowlist path in the
message, that a shifted line stays allowlisted, and that a reasonless entry
is rejected.

* test: touchfile superset check for modules moved out of browse/src/server.ts

Records the paid evals selected by touching browse/src/server.ts at 96764e80
(17 E2E, 1 LLM judge) and asserts every browse/src/routes/*.ts module selects
a superset. The test reads every golden in test/fixtures/moved-module-selection/
so other moved-code goldens can sit beside it.

* test(shard-engine): give non-timeout corpus fixtures CI headroom; keep the POSIX golden off the Windows lane

Only the wall-timeout fixture keeps a 3s wall; the rest get 60s so a loaded
host cannot turn a pass into a timeout. Base and branch runners still agree
on every classification under the new walls. The Windows exclusion entry
moves the free runner's ratchet (c) residual cap to 2356 lines.

* refactor(pty): one runPtySession loop drives observation, counting and floor

test/helpers/pty/session.ts owns launch -> start -> (poll -> tick)* ->
timeout and the failure contract the three runners each hand-rolled: the
run's own error wins over capture and close errors, close always runs, owned
fixture cleanup runs last (also when launch fails). Each runner now supplies a
PtySessionPlan: its boot/command step, poll cadence (2s observation/floor
sleep; counting's output wake + 250ms coalesce), tick policy (permission
handling, native identity, terminal rules stay per runner because they differ)
and capture hooks. The runner bodies are decomposed into top-level steps so no
function exceeds 150 lines; behavior is unchanged and the fake-driver cases
from the first W4 commit pass unmodified.

The counting capture step and the native completion-summary predicate are now
named functions (countingCapture, isNativeCompletionSummary), so
plan-create-prepublication and plan-count-completion call them directly
instead of executing sliced source. The two harnesses that still execute a
sliced runner body with injected boundaries (eng-seeded-completion-ai,
plan-floor-permission) pass the PtyDriver seam instead of overriding
Date/Bun.sleep.

* test(ratchet-c): register route modules, review resolver modules and server.ts residual cap

* refactor(pty): decompose launchClaudePty and engNumberedFindingAUQ under 150 lines

launchClaudePty (349 lines) becomes launch preparation (args, hermetic
child env, owned state roots), recorder creation, spawn, the trust-dialog
watcher, close, and the session handle over one PtyProcess state object. The
failure order is unchanged: abort the viewport, dispose any recorders created
so far, dispose the viewport, rethrow. The --model / --strict-mcp-config
ordering and seedSkills wiring stay pinned by the behavioral fake-CLI tests.

engNumberedFindingAUQ (345 lines) keeps its guards and dispatch; each
self-contained issue family (declared cache, library retry hooks, cache
owner, injected singleton, shared writers, injected export) moves verbatim
into its own function. Every pty/ module is now <= 800 lines and every
top-level function <= 150 lines.

* test(pty): split claude-pty-runner.unit.test.ts along the pty/ module seams

The 188 unit tests move verbatim into claude-pty-runner.{screen,classify,
auq,launch,plan-native,boundaries}.unit.test.ts (test names unchanged; each
file imports only what it uses from the barrel). The five files that no longer
read a SKILL.md template join the test-of-test ratchet baseline with a reason.

* test(touchfiles): moved PTY modules keep their paid-eval selection

test/fixtures/touchfile-selection/w4-pty.json records, at 96764e8, the paid
evals selected by touching test/helpers/claude-pty-runner.ts (20) and
test/helpers/pty-screen.ts (20). touchfiles.test.ts now asserts every .ts file
under test/helpers/pty/ (and pty/screen.ts for both sources) selects a
superset, reading every golden in that directory so later moves can add one;
a planted-violation case pins the report and its Fix line.

* fix(browse): unexchanged pair setup keys no longer authenticate bearer requests

validateToken accepted a gsk_setup_ key as a bearer on /command, /batch and
/file (found while building the W3 auth matrix). A setup key now only
authenticates the /connect exchange.

* W1: one state-root owner (lib/state-root.ts + bin/gstack-state-root.sh), gstack-paths --explain and fail-stop, parity tests

* W1: guarded migration of every executable state-root site; uninstall deletes only ~/.gstack

Bins, careful/freeze hooks, setup, upgrade migrations, browse/src, design,
ios-qa daemon, lib and scripts resolve the state root through
bin/gstack-state-root.sh (bash) or lib/state-root.ts (TS). Bins source the
twin and stop with a reinstall message when it is missing; hooks source it
and never spawn gstack-paths. browse/src/config.ts and lib/cso/state.ts
delegate to resolveStateRoot. Analytics writers and readers move together
so the usage log stays one file. gstack-uninstall deletes state only at
~/.gstack, refuses (exit 2) when it resolves to /, $HOME or an ancestor,
the checkout or the git root, and leaves any other resolved root in place
with the removal command. Fixtures that copy single bins now copy the twin.

* W1: privacy keys and trust-policy deny tiers merge across state roots; gstack-config reporting; test hermeticity

readConfigKey / gstack_read_config_key return the most restrictive
telemetry, memorable_recall, codex_reviews and update_check across the
resolved root and ~/.gstack; other keys read the resolved root only.
gstack-config set reports an overriding root with the exact override
command, list shows the winning root and a root-variable disagreement line.
gstack-gbrain-repo-policy get merges deny/read-only tiers. gstack-egress
reads through readConfigKey. test-setup.ts strips inherited
GSTACK_STATE_ROOT/GSTACK_STATE_DIR and redirects the legacy root.

* W1: shared hook logging helper (hosts/claude/hooks/hook-log.ts)

One hook-errors.log writer: root from resolveStateRoot, 0600 on every
append, opt-in rate limit used only by memorable-user-prompt. The five
hooks route through it.

* W1: docs/state-root.md and README troubleshooting pointer

Precedence table, a real --explain example, the move-your-state recipe,
merged privacy keys, the uninstall rule, the resolver-failure fix, and the
plugin-mode note (evidence gate: no official plugin distribution).

* W1b: template and resolver prose resolve state through guarded gstack-paths; ratchet (a)

Every gstack-paths eval in templates and resolvers carries the fail-stop
guard; executable ~/.gstack paths in bash blocks (context recovery preamble,
eureka log, analytics, project artifacts, upgrade snooze, setup-gbrain lock,
retro snapshots, ship consent marker) use $GSTACK_STATE_ROOT, and the writer
prose that pairs with them points at the printed PROJECT_DIR / RETRO_FILE.
ship drops export GSTACK_STATE_ROOT. SKILL.md regenerated (claude + codex),
ship goldens re-pinned, parity and context-budget caps raised to the measured
sizes with notes. test/state-root-ratchet.test.ts enforces the rule with a
reasoned allowlist; W1 touchfile entries plus a superset golden.

* refactor: apply W1 state-root edits in W2/W3/W5-owned files; one moved-code touchfile golden for all workstreams

* test: fold the moved-code touchfile golden into touchfiles.test.ts; fix integration fixture closure and caps

* v1.91.11.0: CHANGELOG, TODOS, docs and conventions for the refactor wave

* test: re-measure plan-ceo/design-consultation caps and ship goldens after the guarded plan-discovery and spec-review blocks; add the state-root twin to the workflow-boundaries fixture

* fix(windows): migrations resolve their directory with either path separator; state-root parity compares under the HOME Git Bash actually sees

* fix(review,ship): state plan-check timing after smoke expiry and test_stub Skip semantics (review workflow judge clarity)

* test(qa-eval): webhook fix eval asks for the fix loop's post-repair probes; eight-scenario coverage stays in the report-only case and the harness recheck

* test(qa-eval): re-pin the webhook prompt contract to the fix-loop stage; R29 coverage omissions stay bound by the report-only case

* fix(review,ship): plan checks publish a checkpoint before each probe; only the smoke expiry stop is skipped

* fix(qa): carry #2999's checkpoint receipt link, report-template line and full-revision placeholder (identical hunks)

* test(qa-callers): disable git auto maintenance in the caller fixture

Git 2.47+ runs auto maintenance detached after commit; on the CI runner's git
2.55 it rewrote .git/objects fan-out directories while the write observer was
running, which surfaced as unauthorized mutations. Same gc.auto=0 /
maintenance.auto=false guard the shared-libs fixture already uses.

* test(plan-mode-no-op): require prose evidence for the prose-fallback members so a spinner-frame judge verdict cannot end the run as asked

* test(ship-docsync): carry #2999's seeded-attempt docsync harness (identical files)

The doc-sync fault cases replayed attempt 1 before reaching their gate and ran
out of their 285s budget. The fixture now seeds attempt 1 and the parent starts
at the gate under test. Taken byte-identical from origin/capy/audit-fix-wave
(fb526898, e6ac813d, 6ce10ff7, d0c53577, 77cce3be). Local: stale-before,
recovery and late-result 6/6 PASS (97-164s); the whole file 12/12 PASS.
2026-10-01 11:57:48 -07:00

722 lines
26 KiB
TypeScript

/**
* Token registry — per-agent scoped tokens for multi-agent browser access.
*
* Architecture:
* Root token (from server startup) → POST /token → scoped sub-tokens
* POST /connect (setup key exchange) → session token
*
* Token lifecycle:
* createSetupKey() → exchangeSetupKey() → session token (24h default)
* createToken() → direct session token (for CLI/local use)
* revokeToken() → immediate invalidation
* rotateRoot() → new root, all scoped tokens invalidated
*
* Scope categories (derived from commands.ts READ/WRITE/META sets):
* read — snapshot, text, html, links, forms, console, etc.
* write — goto, click, fill, scroll, newtab, etc.
* admin — eval, js, cookies, storage, useragent, state (destructive)
* meta — tab, diff, chain, frame, responsive
*
* Security invariants:
* 1. Only root token can mint sub-tokens (POST /token, POST /connect)
* 2. control scope denied by default — must be explicitly flagged.
* Registry API defaults (createToken/createSetupKey with no scopes)
* stay ['read','write']; the /pair ceremony explicitly grants
* DEFAULT_PAIR_SCOPES (read+write+admin+meta — the pairing ceremony
* is the trust boundary; --restrict narrows, --control must be
* explicit and never rides in via a scopes list)
* 3. chain command scope-checks each subcommand individually
* 4. Root token never in connection strings or pasted instructions
*
* Zero side effects on import. Safe to import from tests.
*/
import * as crypto from 'crypto';
import { READ_COMMANDS, WRITE_COMMANDS, META_COMMANDS } from './commands';
// ─── Scope Definitions ─────────────────────────────────────────
// Derived from commands.ts, but reclassified by actual side effects.
// The key insight (from Codex adversarial review): commands.ts READ_COMMANDS
// includes js/eval/cookies/storage which are actually dangerous. The scope
// model here overrides the commands.ts classification.
/** Commands safe for read-only agents */
export const SCOPE_READ = new Set([
'snapshot', 'text', 'html', 'links', 'forms', 'accessibility',
'console', 'network', 'perf', 'dialog', 'is', 'inspect',
'url', 'tabs', 'status', 'screenshot', 'pdf', 'css', 'attrs',
'media', 'data',
]);
/** Commands that modify page state or navigate */
export const SCOPE_WRITE = new Set([
'goto', 'back', 'forward', 'reload',
'load-html',
'click', 'fill', 'select', 'hover', 'type', 'press', 'scroll', 'wait',
'upload', 'viewport', 'newtab', 'closetab',
'dialog-accept', 'dialog-dismiss',
'download', 'scrape', 'archive',
]);
/** Page-level power tools — JS execution, credential access, page mutations */
export const SCOPE_ADMIN = new Set([
'eval', 'js', 'cookies', 'storage',
'cookie', 'cookie-import', 'cookie-import-browser',
'header', 'useragent',
'style', 'cleanup', 'prettyscreenshot',
]);
/** Browser-wide destructive commands — can kill the server, disconnect headed mode */
export const SCOPE_CONTROL = new Set([
'state', 'handoff', 'resume', 'stop', 'restart', 'connect', 'disconnect',
]);
/** Meta commands — generally safe but some need scope checking */
export const SCOPE_META = new Set([
'tab', 'diff', 'frame', 'responsive', 'snapshot',
'watch', 'inbox', 'focus',
]);
export type ScopeCategory = 'read' | 'write' | 'admin' | 'meta' | 'control';
const SCOPE_MAP: Record<ScopeCategory, Set<string>> = {
read: SCOPE_READ,
write: SCOPE_WRITE,
admin: SCOPE_ADMIN,
control: SCOPE_CONTROL,
meta: SCOPE_META,
};
/**
* Scopes granted by POST /pair when nothing narrower is requested.
* Deliberately full page access (b73f3644 / #907): the trust boundary is the
* pairing ceremony, not the scope. 'control' (browser-wide destructive ops)
* is the only scope that stays opt-in via the control flag. Referenced by
* BOTH server.ts (/pair default) and cli.ts (explicit send) so the two
* defaults cannot silently drift apart again.
*/
export const DEFAULT_PAIR_SCOPES: readonly ScopeCategory[] = ['read', 'write', 'admin', 'meta'];
/**
* Typed error for caller-supplied token options (unknown scope, negative
* rateLimit). HTTP handlers catch it to 400 with the message at the endpoint
* where the typo happened — pre-fix, a bad scope sailed through /pair into a
* poisoned setup key and surfaced as a misleading "Invalid request body" to
* the remote agent at /connect.
*/
export class InvalidScopeError extends Error {}
export function assertValidTokenOptions(scopes: readonly string[], rateLimit: number): void {
const validScopes: ScopeCategory[] = ['read', 'write', 'admin', 'meta', 'control'];
for (const s of scopes) {
if (!validScopes.includes(s as ScopeCategory)) {
throw new InvalidScopeError(`Invalid scope: ${s}. Valid: ${validScopes.join(', ')}`);
}
}
if (rateLimit < 0) throw new InvalidScopeError('rateLimit must be >= 0');
}
/**
* Typed error for a reserved or malformed clientId. `root` is the sentinel that
* checkScope/checkDomain/checkRate and the server command gate use to mean "the
* omnipotent root caller" (validateToken:360), so a scoped token carrying it
* would bypass every enforcement path. Empty/non-string ids collapse distinct
* agents together and break revoke-by-clientId. Request-path writers throw;
* restoreRegistry skips-and-logs so one bad state-file entry can't drop later
* sessions or brick boot.
*/
export class ReservedClientIdError extends Error {}
export function assertValidClientId(clientId: unknown): asserts clientId is string {
if (typeof clientId !== 'string' || clientId.trim() === '') {
throw new ReservedClientIdError('clientId must be a non-empty string');
}
if (clientId.trim().toLowerCase() === 'root') {
throw new ReservedClientIdError("clientId 'root' is reserved");
}
}
// ─── Types ──────────────────────────────────────────────────────
export interface TokenInfo {
token: string;
clientId: string;
type: 'session' | 'setup';
scopes: ScopeCategory[];
domains?: string[]; // glob patterns, e.g. ['*.myapp.com']
tabPolicy: 'own-only' | 'shared';
rateLimit: number; // requests per second (0 = unlimited)
expiresAt: string | null; // ISO8601, null = never
createdAt: string;
usesRemaining?: number; // for setup keys only
issuedSessionToken?: string; // for setup keys: the session token that was issued
commandCount: number; // how many commands have been executed
}
export interface CreateTokenOptions {
clientId: string;
scopes?: ScopeCategory[];
domains?: string[];
tabPolicy?: 'own-only' | 'shared';
rateLimit?: number;
expiresSeconds?: number | null; // null = never, default = 86400 (24h)
}
export interface TokenRegistryState {
agents: Record<string, Omit<TokenInfo, 'commandCount'>>;
}
// ─── Rate Limiter ───────────────────────────────────────────────
interface RateBucket {
count: number;
windowStart: number;
}
const rateBuckets = new Map<string, RateBucket>();
function checkRateLimit(clientId: string, limit: number): { allowed: boolean; retryAfterMs?: number } {
if (limit <= 0) return { allowed: true };
const now = Date.now();
const bucket = rateBuckets.get(clientId);
if (!bucket || now - bucket.windowStart >= 1000) {
rateBuckets.set(clientId, { count: 1, windowStart: now });
return { allowed: true };
}
if (bucket.count >= limit) {
const retryAfterMs = 1000 - (now - bucket.windowStart);
return { allowed: false, retryAfterMs: Math.max(retryAfterMs, 100) };
}
bucket.count++;
return { allowed: true };
}
// ─── Token Registry ─────────────────────────────────────────────
const tokens = new Map<string, TokenInfo>();
let rootToken: string = '';
export function initRegistry(root: string): void {
// Idempotent re-init: same token is a no-op so embedders can call this
// alongside any prior call without fighting. Different token after init
// means a misconfigured caller — throw clearly rather than silently
// invalidate every scoped token already issued.
if (rootToken !== '' && rootToken !== root) {
throw new Error(
'token-registry already initialized with a different token; ' +
'embedders must call buildFetchHandler before any registry-mutating code path'
);
}
rootToken = root;
}
export function getRootToken(): string {
return rootToken;
}
export function isRootToken(token: string): boolean {
// Constant-time compare so a tunnel-reachable caller who can provoke an
// isRootToken() call (e.g., via the 403 "root over tunnel" rejection path)
// can't measure byte-by-byte string-compare timing to recover the token.
// Compare UTF-8 byte lengths (not JS string length) before timingSafeEqual,
// which throws on length-mismatched buffers. A multibyte input whose JS
// string length matches rootToken but whose UTF-8 byte length differs must
// return false on the auth path, not error out.
if (!rootToken) return false;
const tokenBytes = Buffer.byteLength(token, 'utf8');
const rootBytes = Buffer.byteLength(rootToken, 'utf8');
if (tokenBytes !== rootBytes) return false;
const a = Buffer.from(token, 'utf8');
const b = Buffer.from(rootToken, 'utf8');
return crypto.timingSafeEqual(a, b);
}
function generateToken(prefix: string): string {
return `${prefix}${crypto.randomBytes(24).toString('hex')}`;
}
/**
* Create a scoped session token (for direct minting via CLI or /token endpoint).
* Only callable by root token holder.
*/
export function createToken(opts: CreateTokenOptions): TokenInfo {
const {
clientId,
scopes = ['read', 'write'],
domains,
tabPolicy = 'own-only',
rateLimit = 10,
expiresSeconds = 86400, // 24h default
} = opts;
// Validate inputs
assertValidClientId(clientId);
assertValidTokenOptions(scopes, rateLimit);
if (expiresSeconds !== null && expiresSeconds !== undefined && expiresSeconds < 0) {
throw new Error('expiresSeconds must be >= 0 or null');
}
const token = generateToken('gsk_sess_');
const now = new Date();
const expiresAt = expiresSeconds === null
? null
: new Date(now.getTime() + expiresSeconds * 1000).toISOString();
const info: TokenInfo = {
token,
clientId,
type: 'session',
scopes,
domains,
tabPolicy,
rateLimit,
expiresAt,
createdAt: now.toISOString(),
commandCount: 0,
};
// Overwrite if clientId already exists (re-pairing)
// First revoke the old session token (but NOT setup keys — they track their issued session)
for (const [t, existing] of tokens) {
if (existing.clientId === clientId && existing.type === 'session') {
tokens.delete(t);
break;
}
}
tokens.set(token, info);
return info;
}
/**
* Create a one-time setup key for the /pair-agent ceremony.
* Setup keys expire in 5 minutes and can only be exchanged once.
*/
export function createSetupKey(opts: Omit<CreateTokenOptions, 'clientId'> & { clientId?: string }): TokenInfo {
// Only validate when a clientId is supplied; an omitted one gets a safe
// generated `remote-<ts>` default below.
if (opts.clientId !== undefined) assertValidClientId(opts.clientId);
const scopes = opts.scopes || ['read', 'write'];
// ?? not ||: rateLimit 0 is documented as "unlimited" and must survive.
const rateLimit = opts.rateLimit ?? 10;
// Validate HERE, not only at exchange time in createToken — otherwise a
// typo mints a poisoned setup key whose failure surfaces to the wrong
// party (the remote agent, at /connect, as "Invalid request body").
assertValidTokenOptions(scopes, rateLimit);
const token = generateToken('gsk_setup_');
const now = new Date();
const expiresAt = new Date(now.getTime() + 5 * 60 * 1000).toISOString(); // 5 min
const info: TokenInfo = {
token,
clientId: opts.clientId || `remote-${Date.now()}`,
type: 'setup',
scopes,
domains: opts.domains,
tabPolicy: opts.tabPolicy || 'own-only',
rateLimit,
expiresAt,
createdAt: now.toISOString(),
usesRemaining: 1,
commandCount: 0,
};
tokens.set(token, info);
return info;
}
/**
* Exchange a setup key for a session token.
* Idempotent: if the same key is presented again and the prior session
* has 0 commands, returns the same session token (handles tunnel drops).
*/
export function exchangeSetupKey(setupKey: string, sessionExpiresSeconds?: number | null): TokenInfo | null {
const setup = tokens.get(setupKey);
if (!setup) return null;
if (setup.type !== 'setup') return null;
// Check expiry
if (setup.expiresAt && new Date(setup.expiresAt) < new Date()) {
tokens.delete(setupKey);
return null;
}
// Idempotent: if already exchanged but session has 0 commands, return existing
if (setup.usesRemaining === 0) {
if (setup.issuedSessionToken) {
const existing = tokens.get(setup.issuedSessionToken);
if (existing && existing.commandCount === 0) {
return existing;
}
}
return null; // Session used or gone — can't re-issue
}
// Consume the setup key
setup.usesRemaining = 0;
// Create the session token
const session = createToken({
clientId: setup.clientId,
scopes: setup.scopes,
domains: setup.domains,
tabPolicy: setup.tabPolicy,
rateLimit: setup.rateLimit,
expiresSeconds: sessionExpiresSeconds ?? 86400,
});
// Track which session token was issued from this setup key
setup.issuedSessionToken = session.token;
return session;
}
/**
* Validate a token and return its info if valid.
* Returns null for expired, revoked, or unknown tokens, and for unexchanged
* setup keys: a setup key authenticates only the /connect exchange, never a
* bearer request.
* Root token returns a special root info object.
*/
export function validateToken(token: string): TokenInfo | null {
if (isRootToken(token)) {
return {
token: rootToken,
clientId: 'root',
type: 'session',
scopes: ['read', 'write', 'admin', 'meta', 'control'],
tabPolicy: 'shared',
rateLimit: 0, // unlimited
expiresAt: null,
createdAt: '',
commandCount: 0,
};
}
const info = tokens.get(token);
if (!info || info.type !== 'session') return null;
// Check expiry
if (info.expiresAt && new Date(info.expiresAt) < new Date()) {
tokens.delete(token);
return null;
}
return info;
}
/**
* Check if a command is allowed by the token's scopes.
* The `chain` command is special: it's allowed if the token has meta scope,
* but each subcommand within chain must be individually scope-checked.
*/
export function checkScope(info: TokenInfo, command: string): boolean {
if (info.clientId === 'root') return true;
// Special case: chain is in SCOPE_META but requires that the caller
// has scopes covering ALL subcommands. The actual subcommand check
// happens at dispatch time, not here.
if (command === 'chain' && info.scopes.includes('meta')) return true;
for (const scope of info.scopes) {
if (SCOPE_MAP[scope]?.has(command)) return true;
}
return false;
}
/**
* Check if a URL is allowed by the token's domain restrictions.
* Returns true if no domain restrictions, or if the URL matches any glob.
*/
export function checkDomain(info: TokenInfo, url: string): boolean {
if (info.clientId === 'root') return true;
if (!info.domains || info.domains.length === 0) return true;
try {
const parsed = new URL(url);
const hostname = parsed.hostname;
for (const pattern of info.domains) {
if (matchDomainGlob(hostname, pattern)) return true;
}
return false;
} catch {
return false; // Invalid URL — deny
}
}
function matchDomainGlob(hostname: string, pattern: string): boolean {
// Simple glob: *.example.com matches sub.example.com
// Exact: example.com matches example.com only
if (pattern.startsWith('*.')) {
const suffix = pattern.slice(1); // .example.com
return hostname.endsWith(suffix) || hostname === pattern.slice(2);
}
return hostname === pattern;
}
/**
* Check rate limit for a client. Returns { allowed, retryAfterMs? }.
*/
export function checkRate(info: TokenInfo): { allowed: boolean; retryAfterMs?: number } {
if (info.clientId === 'root') return { allowed: true };
return checkRateLimit(info.clientId, info.rateLimit);
}
/**
* Record that a command was executed by this token.
*/
export function recordCommand(token: string): void {
const info = tokens.get(token);
if (info) info.commandCount++;
}
/**
* Revoke ALL tokens for a client ID — the session token and every setup key,
* spent or unspent. Deleting only the first match left two holes: a spent
* setup key (kept for idempotent re-exchange) shadowed the session token, so
* revoke reported success while the live session survived; and an unspent
* setup key surviving revoke let a "revoked" agent POST /connect into a
* fresh session. Returns the number of tokens deleted (0 = nothing found).
*/
export function revokeToken(clientId: string): number {
let deleted = 0;
for (const [token, info] of tokens) {
if (info.clientId === clientId) {
tokens.delete(token); // Map tolerates delete during for...of iteration
deleted++;
}
}
if (deleted > 0) rateBuckets.delete(clientId);
return deleted;
}
/**
* Revoke the PENDING (unspent) setup keys for a client, leaving any live
* session AND spent keys untouched. A re-pair always drops pending keys so a
* superseded broad key can never be exchanged — this closes the shadow-key
* hole (a reducing re-pair before the agent connects would otherwise leave the
* old broad key live) without touching the spent key that #2646 keeps for
* idempotent re-exchange on a tunnel drop. Returns the number deleted.
*/
export function revokeSetupKeys(clientId: string): number {
let deleted = 0;
for (const [token, info] of tokens) {
// usesRemaining !== 0 = still exchangeable (pending). Spent keys (0) are
// harmless: their session is either kept here or revoked on the reduce path.
if (info.clientId === clientId && info.type === 'setup' && info.usesRemaining !== 0) {
tokens.delete(token);
deleted++;
}
}
return deleted;
}
/** The live (non-expired) session token for a client, if any. */
export function getClientSession(clientId: string): TokenInfo | null {
const now = new Date();
for (const info of tokens.values()) {
if (info.clientId !== clientId || info.type !== 'session') continue;
if (info.expiresAt && new Date(info.expiresAt) < now) continue;
return info;
}
return null;
}
/** The effective grant a re-pair is requesting, resolved to concrete values. */
export interface ResolvedGrant {
scopes: ScopeCategory[];
domains?: string[];
rateLimit: number;
tabPolicy: 'own-only' | 'shared';
}
/**
* Does `grant` remove any capability the live `prior` session holds? Drives the
* /pair supersede decision: a reducing re-pair revokes the old session
* immediately (the narrowing must not wait for a reconnect that may never
* happen); a broaden/refresh leaves it working (no outage). Fails toward
* revocation on an unprovable domain superset — a spurious revoke costs one
* reconnect, a missed one leaves wide access live.
*/
export function grantReducesAccess(prior: TokenInfo, grant: ResolvedGrant): boolean {
return scopesReduced(prior.scopes, grant.scopes)
|| domainsReduced(prior.domains, grant.domains)
|| rateReduced(prior.rateLimit, grant.rateLimit)
|| tabPolicyReduced(prior.tabPolicy, grant.tabPolicy);
}
function scopesReduced(prior: ScopeCategory[], next: ScopeCategory[]): boolean {
// Any scope the prior held that the new grant omits (also catches dropping 'control').
return prior.some(s => !next.includes(s));
}
function domainsReduced(prior: string[] | undefined, next: string[] | undefined): boolean {
const priorUnrestricted = !prior || prior.length === 0;
const nextUnrestricted = !next || next.length === 0;
if (priorUnrestricted) return !nextUnrestricted; // universe → restricted = reduce
if (nextUnrestricted) return false; // restricted → universe = broaden
// Both restricted: reduced if any host the prior allowlist admits is no longer
// admitted by the new one. Approximate over patterns — prior is covered iff
// every prior pattern is covered by some next pattern; anything unprovable
// counts as reduced (fail toward revocation).
return prior!.some(p => !next!.some(n => domainGlobCovers(n, p)));
}
/** Does allowlist pattern `wide` admit every host that `narrow` admits? Mirrors
* matchDomainGlob's suffix/exact rules. */
function domainGlobCovers(wide: string, narrow: string): boolean {
if (wide === narrow) return true;
const wideGlob = wide.startsWith('*.');
if (wideGlob) {
const wideSuffix = wide.slice(1); // ".example.com"
const wideApex = wide.slice(2); // "example.com"
if (!narrow.startsWith('*.')) {
// narrow is an exact host; covered iff the wide glob matches it.
return narrow === wideApex || narrow.endsWith(wideSuffix);
}
// narrow is also a glob; its apex must fall under the wide suffix.
const narrowApex = narrow.slice(2);
return narrowApex === wideApex || narrowApex.endsWith(wideSuffix);
}
// wide is an exact host: covers only the identical host (handled by === above).
return false;
}
function rateReduced(prior: number, next: number): boolean {
const priorUnlimited = prior <= 0; // 0 = unlimited
const nextUnlimited = next <= 0;
if (priorUnlimited) return !nextUnlimited; // unlimited → capped = reduce
if (nextUnlimited) return false; // capped → unlimited = broaden
return next < prior; // both capped: a lower cap = reduce
}
function tabPolicyReduced(prior: 'own-only' | 'shared', next: 'own-only' | 'shared'): boolean {
return prior === 'shared' && next === 'own-only';
}
/**
* Rotate the root token. All scoped tokens are invalidated.
* Returns the new root token.
*/
export function rotateRoot(): string {
rootToken = crypto.randomUUID();
tokens.clear();
rateBuckets.clear();
return rootToken;
}
/**
* List all active (non-expired) scoped tokens.
* With includeSetup, unexchanged ("pending") setup keys are listed too —
* they are live grants an operator must be able to see and revoke. Spent
* keys stay hidden: they are re-exchange bookkeeping for a session that is
* already listed.
*/
export function listTokens(opts?: { includeSetup?: boolean }): TokenInfo[] {
const now = new Date();
const result: TokenInfo[] = [];
for (const [token, info] of tokens) {
if (info.expiresAt && new Date(info.expiresAt) < now) {
tokens.delete(token);
continue;
}
if (info.type === 'session') {
result.push(info);
} else if (opts?.includeSetup && info.type === 'setup' && info.usesRemaining !== 0) {
result.push(info);
}
}
return result;
}
/**
* Serialize the token registry for state file persistence.
*/
export function serializeRegistry(): TokenRegistryState {
const agents: TokenRegistryState['agents'] = {};
for (const info of tokens.values()) {
if (info.type === 'session') {
const { commandCount, ...rest } = info;
agents[info.clientId] = rest;
}
}
return { agents };
}
/**
* Restore the token registry from persisted state file data.
*/
export function restoreRegistry(state: TokenRegistryState): void {
tokens.clear();
const now = new Date();
for (const [clientId, data] of Object.entries(state.agents)) {
// Skip expired tokens
if (data.expiresAt && new Date(data.expiresAt) < now) continue;
// Skip-and-log rather than throw: a hand-edited or corrupt state file must
// not brick boot or drop every later valid session. A persisted clientId
// 'root' would otherwise inject a token that bypasses all scope checks.
try {
assertValidClientId(clientId);
} catch (err) {
console.warn(`[browse] restoreRegistry: skipping invalid clientId ${JSON.stringify(clientId)}: ${err instanceof Error ? err.message : String(err)}`);
continue;
}
tokens.set(data.token, {
...data,
clientId,
commandCount: 0,
});
}
}
// ─── Connect endpoint rate limiter (flood protection) ─────
//
// Global-only cap. Setup keys are 24 random bytes (unbruteforceable), so
// rate limiting here is not about preventing key guessing. It caps
// bandwidth, CPU, and log-flood damage from someone who discovered the
// ngrok URL. A legitimate pair-agent session hits /connect once, so
// 300/min is 60x that pattern and never hit accidentally. Per-IP tracking
// was considered and rejected: adds a bounded Map + LRU for defense
// already adequate at the global layer.
let connectAttempts: { ts: number }[] = [];
const CONNECT_RATE_LIMIT = 300; // attempts per minute (~5/sec average)
const CONNECT_WINDOW_MS = 60000;
export function checkConnectRateLimit(): boolean {
const now = Date.now();
connectAttempts = connectAttempts.filter(a => now - a.ts < CONNECT_WINDOW_MS);
if (connectAttempts.length >= CONNECT_RATE_LIMIT) return false;
connectAttempts.push({ ts: now });
return true;
}
// Test-only reset.
export function __resetConnectRateLimit(): void {
connectAttempts = [];
}
// Test-only reset. Zeroes the registry so a subsequent initRegistry call
// always succeeds. Mirrors __resetConnectRateLimit. Needed by tests that
// follow the rotateRoot() pattern — rotateRoot leaves rootToken non-empty,
// which would otherwise trip the initRegistry mismatch guard.
export function __resetRegistry(): void {
rootToken = '';
tokens.clear();
rateBuckets.clear();
}