Put Piggy on Prime Agent, and let it write to the book

Piggy was a hand-rolled OpenAI tool loop. It is now a Prime Agent session —
Prime Intellect's own harness, embedded as a Node library — answering from
PIG's tools and, for the first time, able to put information into the CRM
rather than only read it out.

The harness is a coding agent, so the first job was taking the coding agent
away from it. `noTools: 'all'` plus an explicit allowlist leaves the model
with PIG's ten `pig_*` tools and no bash, no filesystem, no IPython. That
holds under attack: a hostile extension, a skill and a settings file planted
in the agent's own directory, then `setActiveToolsByName` called with every
built-in, still leaves ten tools, all ours. Both lines are load-bearing —
`noTools` alone registers nothing, and the allowlist is what admits our own.

Writing is gated rather than assumed. A change is proposed, not made: the
tool returns a description, the transcript renders a diff card, and nothing
reaches the database until someone presses Apply. Contracts, commitments,
allocations and compliance always stop for a human whatever the mode. Every
write runs through `executeMutation` as the calling user, so their
capabilities and the audit trail apply exactly as they would to a human's.

Four things about the SDK are wrong in its own documentation and cost a
debugging cycle each: models.json does not resolve an env var name for
`apiKey`, it sends the literal string; there is no built-in prime-inference
provider in 0.84.1; a ResourceLoader you pass in is never reloaded for you;
and the stock system prompt is a coding-assistant prompt that must be
replaced — but replacing it also silently removes the tool list, because the
harness only renders that section when it owns the prompt. AGENTS.md records
all four.

The expensive one was thinking level. The harness defaults to `medium`, and
nemotron spent an entire 4,096-token budget reasoning and returned an empty
answer. `low` was worse; `off` omits the parameter so the endpoint's default
wins. An explicit `reasoning_effort: none` via `thinkingLevelMap` took a turn
from 6,195 output tokens to 149.

And a turn is now bounded. The harness loop is `while (true)` with no
iteration cap; a runaway on a frontier model would have eaten the credit it
is supposed to report on. Ceilings on model calls and tokens, enforced both
through the harness hook and independently from the event stream, plus a
per-user daily spend limit — and the ledger now records spend on turns that
fail, which it previously discarded.

Signing in lands on /piggy, which is a workspace: conversations down one
side, the agent in the middle, what it did and what it cost beside it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

apps/piggy/test/config.test.ts

@@ -1,6 +1,6 @@
 import assert from 'node:assert/strict';
 import test from 'node:test';
-import { loadPiggyConfig } from '../src/config';
+import { loadPiggyConfig, loadPiggyTurnLimits } from '../src/config';
 
 const minimum = {
   DATABASE_URL: 'postgres://pig:pig@localhost:54330/pig',
@@ -18,6 +18,53 @@ test('the chat budget is separate from the worker budget, and larger', () => {
   assert.equal(config.PIGGY_MAX_TURNS, 4);
 });
 
+test('a turn has a ceiling on both axes, generous against the measured turn', () => {
+  const config = loadPiggyConfig(minimum);
+
+  // Measured on the live stack against the default model: a one-tool turn is
+  // 2 model calls and 4,922 tokens, a two-tool turn is 3 and 12,265. The
+  // ceilings are roughly three times the busiest of those, which leaves a real
+  // multi-step question room to breathe and still stops a `while (true)` in
+  // seconds rather than in dollars.
+  assert.equal(config.PIGGY_CHAT_MAX_MODEL_CALLS, 8);
+  assert.equal(config.PIGGY_CHAT_MAX_TURN_TOKENS, 40_000);
+  assert.equal(config.PIGGY_CHAT_DAILY_LIMIT_CENTS, 200);
+
+  // PIGGY_MAX_TURNS is the queue worker's own budget and reaches nothing in the
+  // chat path. Keeping them distinct is the point: raising one used to look
+  // like it raised the other, which is how the chat came to have no ceiling at
+  // all.
+  assert.notEqual(config.PIGGY_MAX_TURNS, config.PIGGY_CHAT_MAX_MODEL_CALLS);
+});
+
+test('the ceilings can be read without the rest of the environment', () => {
+  // The chat server is handed a socket and a token and builds the rest from
+  // defaults; it must not start demanding a DATABASE_URL it never uses.
+  assert.deepEqual(loadPiggyTurnLimits({}), {
+    maxModelCalls: 8,
+    maxTurnTokens: 40_000,
+    dailyLimitCents: 200,
+  });
+  assert.deepEqual(
+    loadPiggyTurnLimits({
+      PIGGY_CHAT_MAX_MODEL_CALLS: '3',
+      PIGGY_CHAT_MAX_TURN_TOKENS: '9000',
+      PIGGY_CHAT_DAILY_LIMIT_CENTS: '0',
+    }),
+    { maxModelCalls: 3, maxTurnTokens: 9_000, dailyLimitCents: 0 },
+  );
+  // A ceiling of zero model calls would answer nothing at all, so it is a
+  // configuration error rather than a very strict deployment.
+  assert.throws(
+    () => loadPiggyTurnLimits({ PIGGY_CHAT_MAX_MODEL_CALLS: '0' }),
+    /PIGGY_CHAT_MAX_MODEL_CALLS/,
+  );
+  assert.throws(
+    () => loadPiggyTurnLimits({ PIGGY_CHAT_MAX_TURN_TOKENS: 'plenty' }),
+    /PIGGY_CHAT_MAX_TURN_TOKENS/,
+  );
+});
+
 test('reasoning stays off by default', () => {
   // Reasoning tokens are billed like any other and nemotron-nano's are
   // verbose. The knob exists for debugging, not for the default deployment.