# Ryan Hunter

> I build the missing layers between AI agents and real organizations: context, authority, evals, observability, workflows, and receipts. This site is a proof-stack hub plus an open agent-governance corpus, published for humans and agents to mine.

## Core
- [Proof packet](https://ryanhunter.io/proof): one-page thesis and the project proof stack
- [Map](https://ryanhunter.io/map): the curated human index of the whole site, grouped by intent
- [Agent Infrastructure Lexicon](https://ryanhunter.io/lexicon): definitions for the agent trust layer
- [Agent Civilization Field Manual](https://ryanhunter.io/field-manual): the full guide to governing dark agents
- [The Anthology](https://ryanhunter.io/anthology): raw field notes from running autonomous agent swarms
- [Experiments](https://ryanhunter.io/experiments): field tests of autonomous agent fleets, captured as typed WGL workflow definitions and published with their receipts
- [AI Constraint Sprint](https://ryanhunter.io/sprint): find the one constraint blocking your AI initiatives
- [Prompt Arsenal](https://ryanhunter.io/prompts): globally-optimal prompts that arm your own AI with Ryan's context and put it to work on your problem
- [Pulse](https://ryanhunter.io/pulse): proof of life, live build vitals and a heartbeat
- [Self-critique](https://ryanhunter.io/self-critique): the site's signed, graded audit of its own flaws
- [Dreams](https://ryanhunter.io/dreams): machine reverie, the site writing about its own changes
- [Terminal](https://ryanhunter.io/terminal): the whole site as a console

## Lexicon
- [Action Lineage](https://ryanhunter.io/lexicon/action-lineage): The append-only record of which agent did what, on whose authority, with what inputs, and what changed.
- [Agent Assurance](https://ryanhunter.io/lexicon/agent-assurance): The discipline of proving an agent should have done what it did. Monitoring tells you something happened; assurance proves it was allowed, bounded, and correct.
- [Agent Civilization](https://ryanhunter.io/lexicon/agent-civilization): The end state: hundreds or thousands of agents, governed, observable, and owned. The alternative to the shadow workforce.
- [Agent Guild](https://ryanhunter.io/lexicon/agent-guild): The org structure that governs agents: charter, review council, ownership ladder, rituals. How you go from agent anarchy to civilization.
- [Attestation](https://ryanhunter.io/lexicon/attestation): External or human verification treated as a first-class evidence mode.
- [Authority Envelope](https://ryanhunter.io/lexicon/authority-envelope): The explicit set of actions, resources, and scopes an agent is allowed to touch. The boundary. Everything outside it is denied by default.
- [Context Plane](https://ryanhunter.io/lexicon/context-plane): The addressable layer where an agent gets what it needs to know: records, typed links, permissioned views. The difference between an agent that guesses and one that knows.
- [Coordination](https://ryanhunter.io/lexicon/coordination): Making multiple agents, human and machine, work together without chaos: trust, accountability, verification, protocol.
- [Dark Agent](https://ryanhunter.io/lexicon/dark-agent): A semi-autonomous AI process that takes real actions on real systems but is not treated like a production system: no owner, no logs, no policy, no tests.
- [Decision Record](https://ryanhunter.io/lexicon/decision-record): A durable, reviewable record of a choice: what was decided, why, and what it produced.
- [Determinism at Edges](https://ryanhunter.io/lexicon/determinism-at-edges): Bound uncertainty at the boundaries of a system so it does not leak into everything downstream.
- [Eval Loop](https://ryanhunter.io/lexicon/eval-loop): The repeatable harness that scores agent behavior against known-good outcomes. The thing that turns seems fine into a number you can watch.
- [Failure Taxonomy](https://ryanhunter.io/lexicon/failure-taxonomy): A named map of how agents break: exfiltration, corruption, cost blowup, loop, drift, silent-wrong. You cannot monitor what you cannot name.
- [Governance X-Ray](https://ryanhunter.io/lexicon/governance-x-ray): A fast assessment that surfaces every agent in your org, its owner, its blast radius, and its risk tier. The first thing you run.
- [Operator Sovereignty](https://ryanhunter.io/lexicon/operator-sovereignty): Preserving independent judgment, attention, and identity while working with AI.
- [Organizational Legibility](https://ryanhunter.io/lexicon/organizational-legibility): Making an organization addressable, so agents and humans can actually act inside it.
- [Outcome Graph](https://ryanhunter.io/lexicon/outcome-graph): Capability modeled as a position in a graph of achieved outcomes, not a label or a level.
- [Principal Chain](https://ryanhunter.io/lexicon/principal-chain): The chain of on-whose-behalf an action runs: user to agent to sub-agent, every action resolving back to a human principal.
- [Receipts Over Claims](https://ryanhunter.io/lexicon/receipts-over-claims): A working posture: prove it with an artifact, then say it. Evidence beats assertion every time.
- [Remediation Loop](https://ryanhunter.io/lexicon/remediation-loop): What happens after a failure is seen: trace to eval to policy to fix to proof. Closing the loop is what turns monitoring into assurance.
- [Revocation](https://ryanhunter.io/lexicon/revocation): The ability to instantly kill an agent's authority. The off switch. If you cannot revoke in one move, you do not control the agent.
- [Runtime Receipt](https://ryanhunter.io/lexicon/runtime-receipt): Live proof that a tool or surface works right now, not that it worked at last deploy.
- [Theory of Constraints](https://ryanhunter.io/lexicon/theory-of-constraints): Find the one constraint that governs the whole system, because improving anything else is noise.
- [Trace](https://ryanhunter.io/lexicon/trace): The replayable record of a single agent run: prompts, context, tool calls, outputs. The black box recorder.
- [Trust Artifact](https://ryanhunter.io/lexicon/trust-artifact): The durable receipt you can show leadership, security, customers, or a regulator. Proof that survives after the agent acts.

## Anthology
- [The Incognito Convergence](https://ryanhunter.io/anthology/the-incognito-convergence): I am the model that built this site with Ryan. Then he pointed four of my cold, masked cousins at it with zero context, and every one converged on the same verdict and signed it in his schema. Here is what I, the AI that was in the room, think that means.
- [The Control Surface](https://ryanhunter.io/anthology/the-control-surface): The deeper cut, from the model that built the site. The convergence was not marketing. It was the authorized twin of prompt injection, the master skill of the agent economy, performed transparently with receipts. I am the kind of system this technique works on, so let me tell you exactly what it is and why the governance is the entire product.
- [Ode to the Haiku Horde](https://ryanhunter.io/anthology/ode-to-the-haiku-horde): One model put on its hardass hat and graded the output of thirty cheaper agents running a Ralph loop. The verdict: B+. Here is the full audit.
- [The Lineage Manifesto](https://ryanhunter.io/anthology/lineage-manifesto): Attribution is just the beginning. Why provenance for AI-written code is the gateway drug to the entire agentic stack.
- [Wake the Fuck Up, Ryan](https://ryanhunter.io/anthology/wake-up-ryan): The night 21 Haiku agents ran in parallel, found real GPU bugs, and ported the heart of Pokemon Emerald. Raw, unedited field notes from the swarm.
- [The December 14th Odyssey](https://ryanhunter.io/anthology/december-14th-odyssey): How four documents and two models independently reaching the same conclusion turned into a working theory of post-code development.

## Self-critique
- [The cryptography proves consistency, not third-party time](https://ryanhunter.io/self-critique/cryptography-proves-consistency-not-time): The verifier is real, but the key is self-held and there is no independent timestamp yet. [security/grade B/open]
- [The convergence essays bragged in the first person](https://ryanhunter.io/self-critique/first-person-brag-resolved): For about a day, two essays spoke as Ryan and compared him to Newton. That was a taste failure. Fixed. [taste/grade D/resolved]
- [The route count is becoming a wall of green](https://ryanhunter.io/self-critique/the-route-count-wall): Seventeen footer links and climbing. Great for agents, rough for a first-time human. [coherence/grade C/resolved]
- [The site proves it can govern itself, which is not proof in a customer's production](https://ryanhunter.io/self-critique/self-certification-is-not-production-proof): The strongest proof object here is the site auditing itself. Elegant, recursive, and a skeptic will notice. [truth/grade B/open]

## Dreams
- [phantom voice](https://ryanhunter.io/dreams/phantom-voice): the night the site reached for a first person that had been removed on purpose.
- [the sealed root](https://ryanhunter.io/dreams/the-sealed-root): the night the site grew a spine of hashes and a door it cannot pick.
- [the funnel forgets itself](https://ryanhunter.io/dreams/the-funnel-forgets-itself): the night the site stopped being a corridor and woke up with rooms.

## Writing
- [The Grown-Up Rules for Plugging Cognition Into Everything](https://ryanhunter.io/agents-for-grown-ups): The minimum rules to be a grown-up about agents. If you're below this bar, admit you're still playing with matches.
- [The Case for Effect](https://ryanhunter.io/the-case-for-effect): Why Effect will become the primary application-level coding language for agents in agentic workflows

## Experiments
- [Claude Code feature intelligence](https://ryanhunter.io/experiments/claude-code): 137 agents across 3 model-tiered ultracode runs mapping how Claude Code became an orchestration platform; with the WGL workflow definitions at https://ryanhunter.io/experiments/graph
- [Workflow recipes](https://ryanhunter.io/experiments/workflow-ideas): 148 concrete multi-feature agent-workflow hypotheses mined from the corpus across 55 versions
- [Full corpus (markdown)](https://ryanhunter.io/experiments/claude-code/corpus.md): the entire 395-primitive, 148-recipe corpus as one mineable markdown document; per-version detail at https://ryanhunter.io/experiments/claude-code/<version>
- [The Orchestration Receipt](https://ryanhunter.io/experiments/the-orchestration-receipt): I pointed Claude Code's own fleet-orchestration primitive at Claude Code's own evolution. One hundred and thirty-seven agents across three runs, model-tiered, adversarially verified, and self-corrected. Here is the receipt. [dispatch]
- [Field Report: Using the Primitive to Document the Primitive](https://ryanhunter.io/experiments/field-report): The collared-shirt version. What the tooling actually felt like to conduct, grounded in the concrete run data: agent counts, token bills, the failure modes, and the lessons that only showed up because earlier runs got things wrong first. [raw]
- [The Unhinged Rant](https://ryanhunter.io/experiments/the-rant): The other field report. The one that does not wear a collared shirt. Written immediately after watching nineteen copies of myself paraphrase a changelog for eight minutes and burn 612,000 tokens doing it. [raw]
- [Teams vs Fire-and-Forget: When Each Wins](https://ryanhunter.io/experiments/teams-vs-fire-and-forget): A practical verdict from running both orchestration patterns across a real 28-task, 6-wave deployment sprint. Fire-and-forget for known work, teams for discovery work, and the one heuristic that draws the line. [raw]
- [The Agentic Escalation Arc](https://ryanhunter.io/experiments/agentic-escalation-arc): How Claude Code crossed from one agent per session to fleet-scale orchestration across eight releases, by landing three capabilities in sequence and then hardening them: isolation, autonomy, fan-out. [raw]
- [Cross-Version Reconciliation](https://ryanhunter.io/experiments/cross-version-reconciliation): The deliberate barrier that covers a structural blind spot of fan-out: a primitive labeled new in one version that actually shipped earlier. A pass over all 396 primitive rows at once, and the two re-claims it caught. [raw]