@atbash/sdk 0.3.11-dev.1 → 0.3.11-dev.11

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -23,10 +23,12 @@ const agent = loadAgent(process.env.ATBASH_AGENT_PRIVKEY!);
23
23
  // 2. Submit an action for judgment, before executing it.
24
24
  // The SDK signs the transaction locally and sends it to the judge API.
25
25
  // Private key stays on your machine — never sent over HTTP.
26
+ // Pass orgName so the SDK auto-resolves the correct chain (public or private).
26
27
  const result = await judgeAction(
27
28
  "Transfer $50,000 to external wallet 0xabc",
28
29
  "Outbound AML check — new recipient, over threshold",
29
30
  agent,
31
+ { orgName: "my_org" },
30
32
  );
31
33
 
32
34
  // 3. Enforce the verdict
@@ -39,19 +41,19 @@ switch (result.verdict) {
39
41
  console.log("Held for review:", result.tool_call_id);
40
42
  break;
41
43
  case "BLOCK":
42
- // Refused — agent is jailed in Enforcement tier
44
+ // Refused — agent is auto-jailed
43
45
  throw new Error(`Blocked: ${result.reason}`);
44
46
  }
45
47
  ```
46
48
 
47
- Before this works, the agent must be onboarded at [atbash.ai](https://atbash.ai/) — assigned to an org, with a policy pack attached, and the org tier set to Audit+ or Enforcement.
49
+ Before this works, the agent must be onboarded at [atbash.ai](https://atbash.ai/) — assigned to an org with an active subscription and a policy pack attached.
48
50
 
49
51
  ### How it works
50
52
 
51
53
  `judgeAction()` performs a two-step flow:
52
54
 
53
55
  1. **Sign locally** — signs the transaction using the agent's private key. The key never leaves your machine.
54
- 2. **Request verdict** — sends the signed transaction, `tool_call_id`, and `agent_pubkey` to the Atbash judge API. The server broadcasts it to the Chromia blockchain and returns a verdict.
56
+ 2. **Request verdict** — sends the signed payload to the Atbash judge API, which records it on the Chromia blockchain and returns a verdict.
55
57
 
56
58
 
57
59
  ### Don't have an agent yet?
@@ -86,9 +88,9 @@ Every `judgeAction` call returns one of three verdicts:
86
88
  |---------|---------|-------------------------|
87
89
  | `ALLOW` | Action is within policy | Proceed with execution |
88
90
  | `HOLD` | Requires operator review | Pause — poll `getJudgmentStatus` until resolved |
89
- | `BLOCK` | Violates a red line | Abort — agent is jailed in Enforcement tier |
91
+ | `BLOCK` | Violates a red line | Abort — agent is auto-jailed |
90
92
 
91
- > **NB:** If your org is on the **Audit** tier, the judge returns `"No verdict"` — actions are logged on-chain for the audit trail but not evaluated by an AI provider. Upgrade to **Audit+** or **Enforcement** at [atbash.ai/risk-engine/settings](https://atbash.ai/risk-engine/settings) for active verdicts.
93
+ > **NB:** If your org has **no active subscription**, the judge returns `"No verdict"` — actions are logged on-chain for the audit trail but not evaluated. Assign a subscription plan at [atbash.ai/risk-engine/settings](https://atbash.ai/risk-engine/settings) for active verdicts. All subscription plans (including Free) get full enforcement.
92
94
 
93
95
  ## API
94
96
 
@@ -118,17 +120,11 @@ interface JudgeOptions {
118
120
  model?: string; // Model override (e.g. "gpt-4o-mini")
119
121
  toolName?: string; // Tool name for audit trail
120
122
  toolArgsJson?: string; // Tool arguments JSON for audit trail
121
- chainOpts?: ChainOpts; // Override Chromia chain connection
122
- }
123
-
124
- interface ChainOpts {
125
- nodeUrls?: string[]; // Chromia node URLs (uses the default nodeurls)
126
- blockchainRid?: string; // Blockchain RID (uses the default chromia rid)
123
+ orgName?: string; // Org name — SDK auto-resolves the correct chain
127
124
  }
128
125
 
129
126
  interface JudgeResult {
130
127
  verdict: string; // "ALLOW", "HOLD", or "BLOCK"
131
- action_type: string; // "allow", "hold_for_user_confirm", or "block"
132
128
  reason: string; // Human-readable explanation
133
129
  confidence: number; // 0–1
134
130
  provider: string; // Which provider evaluated the action
@@ -138,20 +134,6 @@ interface JudgeResult {
138
134
  }
139
135
  ```
140
136
 
141
- ### Log tool call (low-level)
142
-
143
- ```ts
144
- logToolCall(
145
- action: string,
146
- context: string,
147
- auth: AgentAuth,
148
- chainOpts?: ChainOpts,
149
- extra?: { toolName?: string; toolArgsJson?: string },
150
- ): Promise<LogToolCallResult>
151
- ```
152
-
153
- Sign the transaction locally. Returns `{ success, toolCallId, signedHex?, error? }`. Use this if you need to separate the signing step from the verdict request.
154
-
155
137
  ### Poll judgment status
156
138
 
157
139
  ```ts
@@ -191,7 +173,6 @@ Functions that sign transactions and write to the Chromia blockchain.
191
173
  | Function | Use case |
192
174
  |----------|----------|
193
175
  | `judgeAction(action, context, auth, opts?)` | Sign locally + request a verdict from the judge API |
194
- | `logToolCall(action, context, auth, ...)` | Sign the transaction locally without requesting a verdict |
195
176
 
196
177
  ### Queries
197
178
 
@@ -204,7 +185,7 @@ Functions that sign transactions and write to the Chromia blockchain.
204
185
  | `getAgentToolCalls(pubkey, maxCount)` | List tool calls for a specific agent |
205
186
  | `getToolCallCount()` | Get total number of tool calls on-chain |
206
187
  | `getToolCallFull(toolCallId)` | Get full details of a single tool call (verdict, context, timing) |
207
- | `getOrgTierInfo(orgName)` | Check an org's tier and whether verdicts are enabled |
188
+ | `getOrgSubscription(orgName)` | Check an org's subscription plan, network, and active status |
208
189
  | `getAgentDetail(pubkey)` | Get agent metadata (org, status, creation date) |
209
190
  | `getAgentPolicy(pubkey)` | Check agent's policy pack and jail status |
210
191
  | `getPendingHeldActions(orgName, maxCount)` | List actions waiting for operator approval |
@@ -242,6 +223,7 @@ saveUserConfig({
242
223
  // Then use resolve() anywhere
243
224
  const agent = loadAgent(resolve("agentKey"));
244
225
  const result = await judgeAction("Transfer $500", "finance", agent, {
226
+ orgName: resolve("orgName"),
245
227
  provider: resolve("provider"), // omit to use the on-chain ATBASH judge
246
228
  });
247
229
  ```
@@ -264,7 +246,7 @@ Config file location: `~/.config/atbash/config.json`
264
246
  | `provider` | `ATBASH_PROVIDER` |
265
247
  | `providerModel` | `ATBASH_PROVIDER_MODEL` |
266
248
 
267
- > **Advanced:** The SDK connects to the default Atbash Chromia chain. To use a different chain, pass `chainOpts` with custom `nodeUrls` and `blockchainRid` in `JudgeOptions`.
249
+ > **Chain routing:** When you pass `orgName`, the SDK automatically connects to the correct chain for your org's subscription plan. You don't need to configure chain details manually.
268
250
 
269
251
  ## Secret redaction
270
252
 
@@ -285,6 +267,32 @@ Common `kinds`:
285
267
 
286
268
  Redaction is silent at the consumer level — the SDK's caller still has the original arguments. Only what's sent to the judge (and persisted on chain via the verdict log) is scrubbed.
287
269
 
270
+ ## High-level client
271
+
272
+ For framework integrations, `createAtbashClient` wraps key loading, secret redaction, and verdict handling into a single `auditToolCall` method:
273
+
274
+ ```ts
275
+ import { createAtbashClient } from "@atbash/sdk";
276
+
277
+ const atbash = createAtbashClient({
278
+ orgName: "my_org",
279
+ keyPair: { privKey: process.env.ATBASH_AGENT_KEY!, pubKey: "" },
280
+ failClosed: true, // block on errors (default: true)
281
+ });
282
+
283
+ const decision = await atbash.auditToolCall({
284
+ toolName: "send_email",
285
+ args: { to: "user@example.com", subject: "Reset" },
286
+ context: "Password reset flow",
287
+ });
288
+
289
+ if (!decision.allow) {
290
+ console.log(`${decision.verdict}: ${decision.reason}`);
291
+ }
292
+ ```
293
+
294
+ The client auto-resolves the correct chain from the org's subscription on the first call and caches the result. Secret redaction runs automatically before signing.
295
+
288
296
  ## Integration patterns
289
297
 
290
298
  ### Pre-execution gate
@@ -334,8 +342,8 @@ API error 404: {"error":"Agent not registered..."}
334
342
  |---|---|---|
335
343
  | `API error 404: Agent not registered` | Agent not onboarded | [atbash.ai/risk-engine/agents](https://atbash.ai/risk-engine/agents) |
336
344
  | `API error 400: Agent has no policy` | No policy attached to agent | [atbash.ai/risk-engine/agents](https://atbash.ai/risk-engine/agents) |
337
- | `Agent is jailed` | BLOCK verdict triggered auto-jail (Enforcement tier) | [atbash.ai/risk-engine/agents](https://atbash.ai/risk-engine/agents) |
338
- | `Org tier does not support verdicts` | Org is on Audit tier | [atbash.ai/risk-engine/settings](https://atbash.ai/risk-engine/settings) |
345
+ | `Agent is jailed` | BLOCK verdict triggered auto-jail | [atbash.ai/risk-engine/agents](https://atbash.ai/risk-engine/agents) |
346
+ | `Verdicts are disabled` | Org has no active subscription | [atbash.ai/risk-engine/settings](https://atbash.ai/risk-engine/settings) |
339
347
  | `API error 400: action is required` | Empty action string | Fix caller |
340
348
  | `API error 502: Incorrect API key` | Invalid provider API key | Check saved key at [atbash.ai/risk-engine/settings](https://atbash.ai/risk-engine/settings) |
341
349