Platform & Open Source

SPECTER

An open-source agent for getting cited by AI search

Role
Author
Year
2026
Status
Open source, MIT licensed
43
skill-graph nodes
5
capability clusters
MIT
license
Weekly
unattended monitoring cadence

The problem

Generative engine optimization was being discussed as an idea and sold as a service, but nobody had written down the operating procedure an agent could follow unattended. A prompt is not a methodology.

What we built

Decisions That Mattered

1.

Write the graph, not the prompt

43 nodes across five clusters, each one loadable on demand. An index node routes the agent to the cluster its current task needs, so a citation-monitoring pass never pays for the scraping architecture it is not using.

2.

Close the loop

Monitor, scrape, pay, generate, publish, verify, repeat. Every stage feeds the next, and verification feeds back into monitoring so the strategy adjusts on evidence.

3.

Give the agent a wallet

Nine nodes cover x402 micropayments, so an agent can pay for metered or paywalled data mid-run instead of stalling at a signup wall.

4.

Run it unattended

A scheduled weekly pass runs the monitoring cluster headless and reports citation gaps with deltas against the previous week. The scheduled version stays report-only: it observes, and a human decides what to publish.

What this demonstrates

Agent-readable methodology: knowledge structured so a model loads only what the current task needs.

Claude Code pluginSkill graphx402Headless CLIlaunchd

Next project

Paredes AI

All work

Get in touch

Start a Project

Tell us what you are trying to build and what is in the way. We will tell you early if we are not the right fit.

starrybodies@gmail.com