Skip to main content

Settings

Color Mode

Theme Skin

Background

Appearance preferences are saved in this browser only.

Environment

Current Environment Production

Built with JEKYLL_ENV=production. Changes require deployment.

Quick Links

Theme & Build

Theme
Jekyll v3.10.0
Last BuildOct 18, 15:39
Repository
ai-world-view/ai-world-view.github.io
Changelog Docs Jekyll

Page Location

Page Info

Layout default
Collection none
Path pages/self-improvement.md
URL /self-improvement/

Source Code

View on GitHub Edit in GitHub.dev
Configuration
Site Config _config.yml

Self-Improvement — the fleet that watches the fleet

The routine AI-agent fleet that reviews, fixes, and evolves the ai-world-view hub and framework — a continuously self-improving system.

Estimated reading time: 9 minutes

Edit on GitHub

Self-Improvement — the fleet that watches the fleet

The hub grows the knowledge bases (AI Orchestration) and spawns new countries on its own. But growing isn’t enough — a system that writes itself also needs to watch and improve itself. This page identifies the routine agent fleet that does that: a meta-layer whose job is the health and continuous improvement of the growth system itself.

The fleet was designed in the reference implementation (the year-of-ai hub, from which this org was planted), and this hub arrived with its safety scaffolding and watchers already built: the kill-switch, the write serializer, the evolution ledger, and the daily watchers all run here today. The framework’s own self-improvement mechanisms — the grow-adjacent prompts (distill, evolve) and skills (learn, pollinate) — exist in lineage/framework/ but are dormant — built for the old per-repo model and not yet wired into the central engine. The fleet below wires those up and fills the remaining gaps, forming a closed loop.

The continuous-improvement loop

OBSERVE ledger- collector DETECT health · pages · secret watch GATE prepublish · fw-review LEARN learn- flywheel ALIGN tick-clock · canon wardens FIX hygiene · rollback EVOLVE tier-ROI · meta-audit GOVERN cost ceiling · kill-switch every cycle feeds the next — learnings make each future tick cheaper & better
Each grow tick feeds telemetry into the loop; what the fleet learns is written back into the canonical framework, so the next tick across all repos improves.

The fleet (25 agents)

Each agent is tagged by origin — net-new (build it), rewire (wire up a dormant framework mechanism), or upgrade (make an existing deterministic tool agentic). Priorities P0 → P2. Agents marked Built run in this hub today; the rest are the designed roadmap.

Monitor / Observe

Agent Pri Origin What it does
telemetry-ledger-collector P0 upgrade Built — appends one record per grow run to the hub evolution ledger (telemetry/evolution.jsonl); the P0 keystone every learn/cost/monitor agent depends on
fleet-pause-killswitch P0 net-new Built — _data/fleet_pause.yml, the single global halt every dispatching/mutating workflow reads first
pages-deploy-sentinel P0 net-new Built — hourly read-only check that each member repo’s Pages build succeeded + the live site responds; files one sticky issue. The sole liveness signal a rollback agent keys off
secret-expiry-watch P0 net-new Built — daily single-attempt probe of CLAUDE_CODE_OAUTH_TOKEN + LIFECYCLE_PAT ahead of the growth window; files/auto-closes one issue (catches a dead token before ticks burn)
fleet-health-watch P1 net-new Built — daily read-only digest over the evolution ledger: error rate, cost trend, per-member last-grow, stalls; one auto-closing issue (the DETECT half of the loop)
fleet-cost-governor P2 net-new Cost rail on the orchestrator — pre-flight ceiling + budget writer
lineage-state-report P2 net-new Single human pane of glass; a pure aggregator over the other signals

Review

Agent Pri Origin What it does
prepublish-gate P0 net-new The biggest hole: an inline gate that builds + fact/link/tone-checks a tick before it publishes
framework-pr-reviewer P0 upgrade Built — safety for the self-modification path; gates every framework PR
injection-surface-auditor P1 net-new Audits the workflow-injection surface (untrusted input into prompts; tools-with-write)
published-content-auditor P2 upgrade Audits published content for post-publish drift — internal-link rot, self-contradiction, orphans
world-view-consistency-auditor P2 net-new Checks that the published sites stay true to the concept: the model’s own world view, with no web-sourced material creeping in

Fix

Agent Pri Origin What it does
repo-write-serializer P0 rewire Built (convention) — the repo-write-<repo> concurrency lock grow-lineage holds + the documented framework/policy single-PR groups; future writers join them so no two race a branch
repo-hygiene-warden P1 net-new Enforces the invariant that a country repo holds only content + config + .claude + telemetry
model-id-drift-checker P1 upgrade Keeps the hardcoded model IDs in policy.yml from going stale or deprecated
supply-chain-security-warden P1 net-new Extends the CodeQL scan (built) with the Dependabot / action-pinning audit on secret-bearing workflows
tick-rollback-sentinel P2 net-new Bounded, human-initiated withdrawal of a bad publish — the rollback the system lacks

Align

Agent Pri Origin What it does
ledger-tickclock-auditor P1 rewire Validates the tick-clock / Evolution Log against reality via a CI check gate

Update

Agent Pri Origin What it does
docs-warden P1 net-new Built — gates every PR + sweeps main so every change is matched by a doc update on the hub’s own doc surface
claude-md-canon-warden P1 rewire Re-syncs each repo’s drifted (stale old-model) CLAUDE.md back to canon
adapter-canon-aligner P2 rewire The backward .claude/-to-canon diff — keeps the framework converging, not fragmenting

Learn

Agent Pri Origin What it does
learn-flywheel P1 rewire Mines a stabilized telemetry window and embeds friction-removing edits into the canonical framework

Evolve

Agent Pri Origin What it does
tier-roi-auditor P2 net-new Makes model-tier selection data-driven instead of fixed-by-intuition
recovery-rehearsal-agent P2 net-new Exercises the re-seed / rebuild path in an isolated scratch org (the untested recovery story)
fleet-meta-auditor P2 net-new Watches the watchers — audits the agent fleet itself, with an external dead-man’s-switch

Defects the design surfaced (fixed here by construction)

The fleet design was run against the reference implementation and surfaced real defects in the old per-repo model. This hub was planted after those findings, so the fixes arrived with it — each one is worth guarding regardless:

  • No evolution ledger — per-tick telemetry used to be discarded after 14 days; here telemetry-ledger.yml appends every run to telemetry/evolution.jsonl.
  • No supply-chain security — here codeql.yml scans the hub; the fuller Dependabot / action-pinning audit remains on the roadmap.
  • A bypassable review — the old learn.yml force-merged its own PR; here framework changes are gated by framework-pr-reviewer.yml.
  • A publish race on each member repo’s main — here every writer joins the repo-write-<repo> concurrency group.
  • A prompt-injection surface — here the engine’s gate job validates dispatch inputs before anything reaches a model prompt.
  • CLAUDE.md drift — member repos can still drift from canon between re-syncs; the canon warden that closes this is on the roadmap.

Rollout order

  1. Phase 0 — safety scaffolding (before any mutating agent): the kill-switch, the write serializer, the gated framework PR review. Planted built.
  2. Phase 1 — keystone + gates + canaries: the telemetry ledger collector, then the secret / Pages / health watchers. Planted built; the inline prepublish gate remains open.
  3. Phase 2 — the loops: the learn flywheel and the align / fix wardens.
  4. Phase 3 — slow structural, economic & meta: tier-ROI, cost governor, content & consistency auditors, recovery rehearsal, and the fleet meta-auditor.

Full detail

The complete fleet — every agent’s trigger, cadence, model tier, inputs, outputs, guardrails, and the closed-loop rationale — is recorded in ADR-0003 in the reference year-of-ai hub, from which this org was planted. It is the synthesized output of a 14-agent design workflow (4 architect lenses → synthesis → 3 adversarial critics → finalize). Or browse the content hub to see the published output those agents protect.