loa-freeside

Single-pass network health observation. Checks API liveness, namespace freshness, drift, computes Network Health Score.

Deep investigation of a single construct. Identity-reality drift, maintenance pattern, composition analysis.

Single-pass network health observation. Checks API liveness, namespace freshness, drift, computes Network Health Score.

Autonomous observation loop. Time-boxed cycles with ratcheting health score, commits findings to grimoires.

Synthesize observations into a readable network health report. Aggregates patterns and trends from observation cycles.

Single-pass coherence sense of an ARBITRARY estate (a registry/state pair). Resolves the read-command from operator-local estate-config, shells the estate's own doctor read-only, emits exactly the STATUS|SIGNAL|MISMATCH tile. Sense-only — never mutates the estate.

Sense whether the config the OPERATOR owns is consistent across an estate of Loa repos. Loa is mounted per-repo, so every repo carries its own .loa.config.yaml and there is NO global tier — operator-owned config (model routing, which models review code, effort) gets duplicated across N repos and DRIFTS. This eye walks the estate, classifies each config key by TIER (operator vs project), and fires `drift` when an operator-tier key — one that SHOULD be identical everywhere — holds more than one distinct value. Optionally flags dead model pins against a model SoT. Emits the STATUS|SIGNAL|MISMATCH tile + a config-drift ledger. Sense-only — never mutates a config, never gates.

Sense the live CONSTRUCT estate — declared ↔ governed ↔ earned (map ↔ installed ↔ done). Replaces hardcoded construct-maps (CLAUDE.md tables, .run/construct-index.yaml, generated adapters) with a fast read-only SENSOR that walks the real packs, the composition declarations, and an earned-authority ledger. Emits the STATUS|SIGNAL|MISMATCH tile + Exhibit A + a three-row composition console. Sense-only — never grants authority, adds a gate, or mutates a pack.

GECKO's 4th eye — the deployment-seam coherence sensor. The first three eyes look INSIDE a construct (identity-reality, composition-authority, runtime-fit); this one looks AT THE SEAM between where the Loa framework PUTS things (the SoT roots in ~/.loa/deployment.yaml) and where each harness consumer LOOKS for them. It senses the manifest-numbness / scattered-symlink / source-vs-installed-lag class: a consumer expecting an artifact at a root the producer placed elsewhere, silently, until a dispatch breaks. Sense-only. CONFLICT (hard, path-provable) vs SMELL (soft). Use when the same deployment incoherence keeps recurring, after a framework/harness layout change, or as a periodic sweep.

Sense when agents use the WRONG-WEIGHT coordination path (reach-in vs spawn-in-cell vs coord) and detect recurring manual cross-cell patterns (desire-paths) that want to become rails. Reads the mutation/audit trail read-only, classifies each cross-cell ACT by weight-used vs weight-needed, and emits a friction report + the STATUS|SIGNAL|MISMATCH coherence tile. DETECTOR-tier: surfaces, never fail-blocks. Sense-only — never mutates any cell.

Sense whether each construct's declared runtime CONTRACT (construct.yaml capabilities + skill frontmatter: model_tier, agent, allowed-tools, downgrade, workflow.gates) coheres with the runtime that actually runs it. Detects capability-reality drift — the #553 agent/write-conflict that silently drops output, model tiers the runtime doesn't offer, write/web denials contradicted by tools, opus pinned for light work, and surfaces gate-owners. Reads the pack territory read-only, emits the STATUS|SIGNAL|MISMATCH tile + a runtime-fit ledger. Sense-only — never mutates a pack, never gates.

Walk the operator's bonfire — local working copies of authored constructs. Classify each by lifecycle glyph (seed/hot/warm/crystallized/cooling/ghost) and emit to the wall. Surface; never decide.

Single-pass network health observation. Checks API liveness, namespace freshness, drift, computes Network Health Score.

Single-pass network health observation. Checks API liveness, namespace freshness, drift, computes Network Health Score.

Convert product changes into grounded community announcements

Research git history to produce structured change timelines

Convert product changes into grounded community announcements

Extract voice profile from existing content into voice.md

Bridgebuilder — Autonomous PR Review

Security and quality audit of application codebase

Execute sprint tasks with production-quality code and tests

Validate sprint implementation against acceptance criteria

Analyze codebase to extract reality into Loa artifacts

Query Flatline protocol knowledge base via NotebookLM integration

Autonomous skill extraction from debugging discoveries. Activates when agents find non-obvious solutions through investigation, experimentation, or trial-and-error. Captures these discoveries as reusable skills for future sessions.

Loa onboarding wizard: environment detection, profile questionnaire, config generation, and post-config explanation in ≤10 turns

Pre-install / pre-publish manifest linter for construct packs. Emits Verdict stream rows on findings. Checks required manifest fields, path resolution, route declarations, and the CLAUDE.md grimoires-section convention.

Red Team — Generative Adversarial Security Design

Create comprehensive Software Design Document based on PRD

Autonomous agent execution mode

Browse and install construct packs from the Loa Constructs Registry

Triage a bug report through structured phases and create micro-sprint

BUTTERFREEZONE Generation Skill

L2 cost-budget enforcer — daily token cap with fail-closed semantics under uncertainty (billing-API primary, internal counter fallback, periodic reconciliation cron)

L5 cross-repo status reader — reads structured cross-repo state via gh API with TTL cache + stale fallback, BLOCKER extraction from NOTES.md, per-source error capture, p95 <30s for 10 repos

Design and deploy production infrastructure

Launch PRD discovery with codebase grounding and context ingestion

Enhance a prompt for better outputs with analysis and optimization

Run evaluation suites against the Loa framework

L4 graduated-trust ledger — per-(scope, capability, actor) trust tier with hash-chained immutable history, configurable transitions, auto-drop on operator override + cooldown enforcement, and force-grant audit-logged exceptions