A developer agent OS
Cognethics ships the PJ Agent Orchestration System — a set of MCP primitives for spawning your own agents under personas, with explicit USD budgets, persistent memory, hierarchical delegation, and a tamper-evident audit chain. Same tenant, same data, same RBAC as the rest of the platform. Any authenticated tenant user with the right per-tool permission can call them.
How it composes
- Missions and budgets. A mission bounds a goal and a dollar budget; every action an agent takes is attributable to its mission, and spend is checked before each step.
- Spawn and delegate. An agent is created under a named person's authority with that person's permissions; anything beyond them needs a time-limited grant a named approver signed off. It can delegate a task to another agent and escalate anything above its authority.
- Approval before consequence. Consequential actions stop in an approval queue; a human approves or rejects, and the exchange appears in the audit chain.
- Pause, resume, hand off. A long-running agent can be paused and resumed in place, or handed off across instances without losing context.
Production evidence — an accounts-payable validation mission
Parent agent intakes invoices, child validator agent verifies them, escalates above its authority where required, parent approves, child retires. This mission was orchestrated through the PJ Agent Orchestration System, with a budget cap enforced before every step and the exchange written to the audit chain.
- Mission cost: $6.13 against a $10 hard budget — 61% used.
- Cache hit rate: 94.3% across 9 M cache-read tokens.
- Spawn-to-retire wall-clock: ~10 minutes, end to end.
- Audit chain: 9 unbroken hash-linked rows for the parent.
These figures are measured on the reference accounts-payable mission. Costs are reported by the underlying model provider; cache-hit rate and audit chain integrity are measured in-platform.
Same permissions, same data, same observability
- Same permissions. Whether a call comes from a PJ-orchestrated agent or a human, it hits the same dispatcher and the same RBAC check. Per-org scope catalogs at
/.well-known/oauth-authorization-server/groups/{org}/apply identically. - Same data. Agents read and write the multi-tenant business database — the same one that powers the 62 applications. No separate "agent data plane".
- Same observability. Per-agent cost telemetry rolls into the mission view; the audit chain returns ordered, hash-verified rows for any mission.