claude
f0173440e4
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>
36 lines
1.7 KiB
YAML
36 lines
1.7 KiB
YAML
# pnpm workspace definition.
|
|
#
|
|
# Replaces npm's `workspaces` field, which has been removed from package.json —
|
|
# pnpm ignores that field, and keeping both would leave two sources of truth
|
|
# that can silently disagree.
|
|
packages:
|
|
- 'packages/*'
|
|
- 'apps/*'
|
|
|
|
# Dependency install scripts.
|
|
#
|
|
# pnpm refuses to run them unless a package is named here, and treats anything
|
|
# undeclared as a hard error on *every* command until it is resolved. That is a
|
|
# real supply-chain protection: a transitive dependency cannot execute code at
|
|
# install time simply by being pulled in.
|
|
#
|
|
# `esbuild` asks to run one, and does not need to. Its platform binary arrives
|
|
# through an optional dependency (`@esbuild/linux-x64`); the postinstall only
|
|
# verifies what is already present. Checked rather than assumed — after a
|
|
# scriptless install the binary reported 0.25.12 and both the Vite build and
|
|
# tsx work normally.
|
|
#
|
|
# So it is denied rather than approved: same result, one fewer package allowed
|
|
# to execute code on every machine that installs. If a build ever genuinely
|
|
# fails for want of a script, flip it to `true` here and record why.
|
|
# `@google/genai` and `protobufjs` arrive transitively under the Prime Agent SDK,
|
|
# which carries a client for every provider it supports whether or not we use it.
|
|
# Piggy talks to exactly one — Prime Inference, over the OpenAI-compatible API —
|
|
# so neither package is on any path we execute, and their install scripts are
|
|
# denied on the same reasoning as esbuild's. Verified after a scriptless install:
|
|
# the agent session starts, resolves the model and completes a tool round trip.
|
|
allowBuilds:
|
|
'@google/genai': false
|
|
esbuild: false
|
|
protobufjs: false
|