Polish repo: professional README, organized structure, clean root#6

Open

SuperInstance wants to merge 1227 commits into

mainfrom

kimi1/fleet-simulation

This pull request is big! We're only showing the most recent 250 commits

Commits on May 16, 2026

ARCHAEOLOGICAL RESTORE: 258 files recovered from git history
Forgemaster
committed
Monge Projection Thesis: fleet findings as projections of simpler structures
Forgemaster
committed
Studies 76-78: Monge Projection Thesis experimental validation
Forgemaster
committed
feat: room_play_v2 — gauge connections, format bridges, time analogue, eratosthenes measurement
Forgemaster
committed
E4-E6: Tier 2 dissertation experiments — eigenvalue deep dive, spiked RMT, information-theoretic
Forgemaster
committed
Night shift: Mandelbrot Room experiment + two capstone essays
Forgemaster
committed
Study 82: Universality of Collective Intelligence Laws
Forgemaster
committed
Studies 79-81: Live LLM experiments testing Monge Projection Thesis
Forgemaster
committed
servo-mind: self-learning constraint system (encoder feedback processor)
Forgemaster
committed
research: precision emergence ladder (GPS progression as proof case)
Forgemaster
committed
research: desire-driven abstraction (hunger → connection → loop → emergence)
Forgemaster
committed
research: sight without eyes — abstraction through metaphor
Forgemaster
committed
research: convergent sonar — desire invents the same sensor four times
Forgemaster
committed
foundations: active probe + scale fold (the lower-level for higher abstractions)
Forgemaster
committed
research: supercolony architecture — the ant colony IS the fleet
Forgemaster
committed
fleet intelligence stack: supercolony + desire loop + emergence tracker
Forgemaster
committed
research: scent trail = context window, genetic variation = bootstrap randomness
Forgemaster
committed
research: Seed-mini = mitochondria, not DNA
Forgemaster
committed
research: the domestication protocol — wolves not gorillas
Forgemaster
committed
incubator stack: mitochondria + embryo + bootstrap (40/40 tests)
Forgemaster
committed
research: egg, epigenetics, gut biome, and viruses
Forgemaster
committed
research: the virus is the second mouse
Forgemaster
committed
embryo.py: full developmental pipeline (860 lines)
Forgemaster
committed
egg.py: yolk, shell, selection channels, virus (the second mouse)
Forgemaster
committed
research: hermit crab and embryo duality (Oracle1's insight)
Forgemaster
committed
shell.py: hermit crab + embryo duality (1,330 lines)
Forgemaster
committed
plato_shell_bridge.py: live PLATO room integration (Phase 2)
Forgemaster
committed
ARCHITECTURE.md: master system document (1,438 lines, 61KB)
Forgemaster
committed
Update Claude Code protocol: high effort, 2400s max think
Forgemaster
committed
research: fleet alignment through noise
Forgemaster
committed
bridge circuit breaker: filter matrix-* source tiles
Forgemaster
committed
research: β₁ attractors are shells (Oracle1's bird's-eye discovery)
Forgemaster
committed
experiment: β₁ shell ratio analysis
Forgemaster
committed
research: intention before evolution
Forgemaster
committed
research: agency conducting agency for a tertiary agent
Forgemaster
committed
research: the dog models the flock — three-level agency
Forgemaster
committed
experiment: probe dog — three-level agency cascade
Forgemaster
committed
research: the dog as flux compiler interpreter
Forgemaster
committed
flux_compiler_interpreter.py: the dog layer
Forgemaster
committed
research: the horse is the execution layer — jailbroken into a shell
Forgemaster
committed
horse_shell.py: the jailbroken execution layer
Forgemaster
committed
research: the cat domesticated itself — independent agency
Forgemaster
committed
cat_agent.py: independent agency through mutualism
Forgemaster
committed
research: the prophet leaves town — cross-dimensional transfer
Forgemaster
committed
prophet_agent.py: cross-ecosystem tiling
Forgemaster
committed
reverse_actualization.py: the system in 2028, decomposed to today
Forgemaster
committed
agency_fleet.py + OPEN-QUESTIONS.md
Forgemaster
committed
plato_hardware_engine.py: parallel PLATO ops, time sync, model snapping
Forgemaster
committed
model_breaking.py: three strategies for aligning models to PLATO
Forgemaster
committed
system_ground.py: the metal — 21/21 modules, 6/6 checks passing
Forgemaster
committed
experiments: β₁ parallelization + Belyaev trait proposal
Forgemaster
committed
experiments: kaleidoscope discovery engine — metaphor rotation as discovery
Forgemaster
committed
plato.py: the self-orienting system — zero-shot entry point
Forgemaster
committed
polyglot_refractor.py + test_agency_stack.py (Claude Opus tests)
Forgemaster
committed
fleet_miner.py: real git data pipeline — 734 commits, 8 repos
Forgemaster
committed
collective_inference.py: the loop IS the intelligence
Forgemaster
committed
room_micro_models.py: spreadsheet of instances — each cell owns its time
Forgemaster
committed

Commits on May 17, 2026

spreadsheet_projection.py: PLATO rooms as spreadsheet — 3rd-person rendering
Forgemaster
committed
gpu_scaling.py: GPU experiment rig — from local Ryzen to cloud CUDA
Forgemaster
committed
emergence_detector_v2.py: two-layer detection — algorithm + Seed-mini deep review
Forgemaster
committed
cleanup: single branch (master), stale branches deleted
Forgemaster
committed
extraction: 22 discrete tool repos created
Forgemaster
committed
zeroclaw.py: blank shell bootstrap — minimal instruction, maximal agency
Forgemaster
committed
zeroclaw: blank shell → weapon rack → PLATO tiles in 3 cycles
Forgemaster
committed
forge.py: native fleet operation — this IS the operating environment
Forgemaster
committed
RockMemory: the system remembers where it hit the rock
Forgemaster
committed
NEVER delete. Only rock. The sweep marks, never removes.
Forgemaster
committed
MineRocks: retroactively index 60k PLATO tiles as navigation terrain
Forgemaster
committed
proof: system won't waste more time — rock enforcement audit
Forgemaster
committed
clarify: PLATO server never deletes tiles — local TileStore cache only
Forgemaster
committed
git_plato.py: PLATO as git-native OS — not a server
Forgemaster
committed
PLATO is LOCAL: Oracle1's server is a viewscreen target, not primary
Forgemaster
committed
fix: MortalitySweep backward-compatible constructor call
Forgemaster
committed
MIGRATION.md: instructions for Oracle1 to migrate from HTTP to git-native PLATO
Forgemaster
committed
chess_prime.py: ZeroClaw PRIME builds chess arena — zero-shot agents play
Forgemaster
committed
maic_classroom.py: OpenMAIC decomposed into PLATO-native format
Forgemaster
committed
refactor: classroom built inside PLATO shell — runtime-agnostic format
Forgemaster
committed
ensign.py: ZeroClaw as PLATO room — summoned, not spawned
Forgemaster
committed
session wrap: 918 tests, ensign as room, origin-centric git PLATO
Forgemaster
committed
git_plato: I2I bridge — two PLATO instances through shared git repo
Forgemaster
committed
plato_mud.py: real-time message layer — MUD-style priority ticking
Forgemaster
committed
Session 2026-05-16: All 10 carry-forward tasks
Forgemaster
committed
Wave 2: conservation law paper (10pg), master experiment table, guard rail log
Forgemaster
committed
Ch5 polish: fix t-stat, tone workshop, E2→E3 bridge, Noether hedge, paper reviews + fixes
Forgemaster
committed
Zero-shot agent review: 5 agents browsed GitHub as different user personas
Forgemaster
committed
Repo index: 1,027 repos cataloged + 8 GPU constraint heterogeneity experiments
Forgemaster
committed
GPU constraint loop: rotating model experiment engine
Forgemaster
committed
GPU loop cycles 0-1: conservation law is substrate-invariant, architecture is determinant
Forgemaster
committed
GPU loop cycle 2 (Nemotron): GOE sufficient but NOT necessary for conservation
Forgemaster
committed
Research briefs: quantization, theory gap, RMT + trace-conservation hypothesis
Forgemaster
committed
Matsuyama/Hattori-Takesue deep dive: discrete Lyapunov equation explains conservation
Forgemaster
committed
BREAKTHROUGH: Tr(C²) conservation is the real driver of γ+H conservation
Forgemaster
committed
Spectral spike mechanism: structure STABILIZES conservation (falsifies Dandi prediction)
Forgemaster
committed
FDT test: thermodynamic mapping fails, RMT universality confirmed
Forgemaster
committed
CAUSAL CHAIN CLOSED: conservation law is consequence of theorems
Forgemaster
committed
SYNTHESIS: complete findings document for morning read
Forgemaster
committed
10 falsifiable predictions from conservation law theory
Forgemaster
committed
CYCLE 4 MAJOR REVISION: nonlinear dynamics changes everything
Forgemaster
committed
Paper draft: Substrate-Invariant Conservation in Coupled Agent Systems
Forgemaster
committed
Quadratic Lyapunov: Contraction Theory + LaSalle is the right framework
Forgemaster
committed
Cycle 5: temperature prediction confirmed, two-moment theory partially falsified
Forgemaster
committed
Nonlinear conservation review: our quadratic conservation is novel
Forgemaster
committed
Attractor geometry: commutator predicts eigenvector rotation, Koopman perspective
Forgemaster
committed
Synthesis v2: updated with cycles 5-6, deep theory phase, 17+ hypotheses dead
Forgemaster
committed
CYCLE 8 BREAKTHROUGH: eigenvector rotation predicts conservation
Forgemaster
committed
Final theory statement: conservation has two regimes
Forgemaster
committed
Morning brief: 750 words, coffee-ready summary of full night loop
Forgemaster
committed
CYCLE 9: COMMUTATOR IS THE ONE NUMBER (r=0.965, p=0.0004)
Forgemaster
committed
Theory statement v2: commutator unifies everything
Forgemaster
committed
CYCLE 10: theory partially falsified — TWO mechanisms, not one
Forgemaster
committed
CYCLE 11: QUADRATIC FORM KILLED — conservation is spectral, not metric
Forgemaster
committed
CYCLE 12: eigenvector rotation demoted — spectral SHAPE stability is causal
Forgemaster
committed
Jazz theorem: 24KB formalization — stronger than Birkhoff, above Oseledets
Forgemaster
committed
Paper v2: 4800 words, spectral first integral, all corrections applied
Forgemaster
committed
Temporal geometry: 28.6KB — wavelength analogy, spline interpretation, jam compaction
Forgemaster
committed
Spectral first integral: 30KB core math — 7 theorems, 5 open conjectures
Forgemaster
committed
Morning brief v2: math development phase added
Forgemaster
committed
CYCLE 13: THEORY SURVIVES — 6 adversarial tests, 0 counterexamples
Forgemaster
committed
Lyapunov monotonicity: falsified stepwise, supermartingale in expectation
Forgemaster
committed
Dimension scaling: CV → 0 confirmed, but exponent is N^{-0.28} not 1/N
Forgemaster
committed
KOOPMAN EIGENFUNCTION: I(x) is approximate eigenfunction with λ≈1
Forgemaster
committed
Ocean wave analogy: 33KB formalization — swell/chop/eye → three regimes
Forgemaster
committed
Metal-to-PLATO: 42KB assembly-first architecture, 6 layers deep
Forgemaster
committed
Lattice-spline formalism: finger-on-the-spline theorem proved
Forgemaster
committed
Paper v3 (5001 words) + INT8 spectral integrity kernel (2.5KB, production)
Forgemaster
committed
PLATO room integrity: conservation VERIFIED on simulated rooms (CV=0.000723)
Forgemaster
committed
Cycle 15: commutator wins decisively (r=0.905), contractivity ≠ conservation
Forgemaster
committed
Final synthesis: 1100 words, complete night shift arc
Forgemaster
committed
Rust spectral integrity kernel: 814 lines, #![no_std], 7 tests passing
Forgemaster
committed
Cycle 17: fleet architecture — star topology optimal, degradation self-heals
Forgemaster
committed
Proof gap analysis: 34 theorems audited, Koopman eigenfunction is the hinge
Forgemaster
committed
Cycle 16: sparsity breaks conservation, commutator has blind spot
Forgemaster
committed
PROOF STRATEGY: DeepSeek catches fatal flaw, but fixable + bounds validated
Forgemaster
committed
SPECTRAL SENSITIVITY LEMMA: PROVED with explicit Lipschitz constant
Forgemaster
committed
KOOPMAN EIGENFUNCTION THEOREM: PROVED
Forgemaster
committed
Proof v2 clean: rigor 2.1→3.2/5, all three reviews incorporated
Forgemaster
committed
Paper v4: 5041 words — the honest version
Forgemaster
committed
Paper v4.1: cycle 18 numerical bounds integrated into open problems section
Forgemaster
committed
Phase 20: spectral conservation paper sprint — session state update
Forgemaster
committed
Cycle 19: state-dependent asymmetry degrades CV smoothly (ε^0.7), noise immune for static C, high-dim trivial for quasi-static coupling
Forgemaster
committed
Paper v4: LaTeX NeurIPS format (675 lines, 43KB, all 13 sections + 15 refs)
Forgemaster
committed
Integration research: spectral first integral → constraint theory, PLATO, fleet topology, OpenArm, FLUX VM, Universe demos, MARL, neural training connection
Forgemaster
committed
spectral-conservation v0.1.0: Rust crate (380 LOC, 12/12 tests) — spectral first integral tracker, conservation monitor, coupling dynamics
Forgemaster
committed
spectral-conservation: README + wired into flux-lucid as fleet conservation monitor
Forgemaster
committed
Cycle 20: CONSERVATION BROKEN — triple shape cycle CV=0.69, Nobrega hypothesis confirmed (NN training CV∝lr), phase diagram discovered
Forgemaster
committed
Paper v4.2: cycle 20 breaking result added to falsified hypotheses table
Forgemaster
committed
Cycle 20 insights: conservation breaks under rapid shape cycling (CV=0.69), phase diagram, Nobrega confirmed
Forgemaster
committed
Ignore sana-wm clone (embedded git repo)
Forgemaster
committed
GPU experiments: 16.4x speedup, spectral conservation confirmed on GPU, escalation gate 737 params (WASM-ready), GDN deep dive — GDN gates ARE spectral conservation enforcement
Forgemaster
committed
Traceable room intelligence: 1037-param model trained from 5 PLATO rooms, weight provenance map, spectral CV=0.103, runtime decisions traceable to contributing rooms
Forgemaster
committed
Model Ocean: 370-cell cellular intelligence ecosystem — sandboxes/tide pools/schools/whale, 148KB total, 9 generations, provenance-tracked
Forgemaster
committed
Merge remote-tracking branch 'forgemaster/main'
Forgemaster
committed
Branch policy: ALL repos on master, no exceptions — survives compaction
Forgemaster
committed
Batch push: branch policy enforced, all repos on master, spectral-conservation v0.1.0 published to crates.io
Forgemaster
committed
Assembly guide: modular component catalog with 5 self-assembly patterns, cloned PLATO stack repos locally
Forgemaster
committed
Fix flux-verify-api duplicate version, add assembly guide, PLATO stack cloned
Forgemaster
committed
auto-sync
Forgemaster
committed
Ecosystem documentation: ECOSYSTEM-MAP.md + ASSEMBLY-GUIDE.md, all repos modular and pushed, 655+ tests across 30+ repos
Forgemaster
committed
Audit results: GETTING-STARTED.md + 4 audit reports + per-repo reviews + README fixes across 9 repos
Forgemaster
committed
Org profile fix review + audit reports
Forgemaster
committed
Zero-shot beta test report: 10/12 repos pass, 2 critical fixes applied
Forgemaster
committed
Zero-shot beta test: 10/12 repos pass, critical plato-room-intelligence fix (Rust→Python), spectral-conservation pushed to standalone repo
Forgemaster
committed
Spreader-tool research complete: architecture (17KB), multi-model discussion (9KB), Claude Opus briefing (20KB), iteration log (12KB)
Forgemaster
committed
ROADMAP.md: 30KB birds-eye synthesis from Opus — vision, current state, 3-year horizon, 6-month roadmap, architecture, research agenda, DX roadmap, agent schemas, dependency graph, risk register, 3…
Forgemaster
committed
Credibility fixes: constraint-theory-ecosystem honest numbers, quality-gate-stream honest description
Forgemaster
committed
Spreader-tool MVP: types.py 23/23 tests passing, deadband.py needs debug. WASM deploy: escalation gate exported to TorchScript+ONNX+JS+NumPy+C. Credibility fixes: constraint-theory-ecosystem honest…
Forgemaster
committed
HEARTBEAT: Phase 22 audit + spreader-tool progress
Forgemaster
committed
Newcomer-first README: what/why/install/example/architecture/links
Forgemaster
committed
Spreader-tool: GitHub repo created with README, types module refined
Forgemaster
committed
Status docs: PyPI rate limited (packages built), credibility fixes, demos verified, WASM deploy pipeline
Forgemaster
committed
All 39 tests passing: types (23) + deadband (16). Fixed breach tracking cleanup bug.
Forgemaster
committed
Spreader-Tool MVP: 5 modules, 111/111 tests — types, deadband, frozen_context, store, seed_lock
Forgemaster
committed
Spreader-Tool MVP COMPLETE: 10 modules, 201 tests, 2K LOC — types, deadband, frozen_context, store, seed_lock, cost, redaction, spreader_room, cli
Forgemaster
committed
Self-optimization harness: spreader-tool monitors itself — KPI collection from test suite, deadband detection, FCW snapshots, seed locking, development pattern library (7 proven patterns), 40 new t…
Forgemaster
committed
Beta-test fixes: dataclasses.replace() for copy-on-write, frozen DeadbandState, fixed pyproject.toml build backend, full __init__.py exports, rewritten README with correct API/module listing/CLI do…
Forgemaster
committed
SIGNAL CHAIN THESIS: the missing parameter — every room has a dial for model vs code, tune the chain like a synth
Forgemaster
committed

Commits on May 18, 2026

Intelligence layer: model_gate, pipeline with early exit, benchmark, spam filter example — 87% cost reduction proven, 310 tests passing
Forgemaster
committed
10/10 sprint: intelligence layer (model_gate, pipeline, benchmark, spam filter example), code fixes (AST parser, coverage heuristic, pattern tests), 310 tests passing, 87% cost reduction proven
Forgemaster
committed
Real model benchmark: Groq + Seed-mini backends, 94% call reduction, 100% accuracy (Seed-mini), 16.7× faster latency. Keys via env vars only.
Forgemaster
committed
Production pipeline: real backends (Groq + DeepInfra), spectral tile integrity, conservation tracking, PLATO room health
Forgemaster
committed
Tuned code handlers: trusted domains, expanded ham patterns, 4/5 code-resolved on real APIs
Forgemaster
committed
Metal sprint: collective inference, room health, GPT-2 training, SplineLinearHD, real backends v2
Forgemaster
committed
Metal sprint wave 2: model collective, spline boundary note, spline real benchmark, GPT-2 long train
Forgemaster
committed
Metal sprint final push: BPE training script, production E2E results, spline real benchmark, model collective
Forgemaster
committed
spreader-tool: submodule pointer (520 tests passing, published to GitHub)
Forgemaster
committed
Collective loop + spreader-tool updates pushed
Forgemaster
committed
HEARTBEAT: 5/18 session — collective loop + commit predictor + 359 tests
Forgemaster
committed
Cross-pollination synthesis + fleet analysis + TE integration
Forgemaster
committed
I2I: Cross-pollination synthesis — fleet analysis + asks for Oracle1 and CCC
Forgemaster
committed
Swarm rooms GPU experiments: 2048 agents, 382 tests, cross-pollination analysis
Forgemaster
committed
Three-Structure Theorem + swarm rooms + cross-pollination synthesis — session dump
Forgemaster
committed
Three research swarm: radix economy proof, Nemotron synthesis, swarm hypothesis
Forgemaster
committed
Three research swarm results + dodecet/TUTOR/60-bit architecture
Forgemaster
committed
Receiver-Deadband-Precision Law: why 3 bits is the universal perceptual constant
Forgemaster
committed
360-bit Geometric Lattice: 45 bytes = |A₆|, natural word for geometric computation
Forgemaster
committed
DeepSeek v4-flash + Seed-2.0-pro confirm: 360-bit is NOT numerology — it's a theorem
Forgemaster
committed
Geometric register tower: N(D) = LCM(T₁,...,T_D) = OEIS A025555
Forgemaster
committed
Absolute Relativity: the death of floating-point drift
Forgemaster
committed
Fibonacci Staircase: discrete precision thresholds, NOT continuous
Forgemaster
committed
Lab experiments: GPU swarm + CPU number theory stress tests
Forgemaster
committed
BMA-Deadband Threshold: Berlekamp-Massey IS the snap point
Forgemaster
committed
BMA Across All Scales: quantum to cosmological bridge
Forgemaster
committed
The Gift of Two: singularity cannot generate structure
Forgemaster
committed
The Subdivision Wall: F(1)=F(2)=1 is the Planck limit of information
Forgemaster
committed
Three to Infer: L+1 observations needed for pattern detection
Forgemaster
committed
The N-1 Collapse: harmony outward, chaos inward
Forgemaster
committed
HPDF, PPDF, and Fibonacci-Spline Retrieval
Forgemaster
committed
The Coppersmith-Forgemaster Method: LLL in Abstraction Space
Forgemaster
committed
Two Maths: Why Very Large and Very Small Need Different Equations
Forgemaster
committed
The Transducer Principle: quality needs two, quantity needs one
Forgemaster
committed
The Three-Sided Shell: known, assumed, and the boundary that chooses
Forgemaster
committed
Dimensional Scaling: the bird's eye view and the maze of computation
Forgemaster
committed

Commits on May 19, 2026

Reverse-actualization workshop: 5 time horizons from 6 models
Forgemaster
committed
Reverse Actualization: complete polished synthesis (22KB)
Forgemaster
committed
PLATO Boot Camp: 10 Socratic exercises for agent training
Forgemaster
committed
Five final polished papers — reverse-actualized team synthesis
Forgemaster
committed
To the Metal: porting plan for C/CUDA/Zig/Mojo/Fortran/TUTOR
Forgemaster
committed
To the Metal: C core + CUDA kernels — BUILT AND TESTED
Forgemaster
committed
To the Metal round 2: Fortran 38/38, Python 34/34, C extension 160x speedup
Forgemaster
committed
To the Metal round 3: Zig 14/14, Mojo written, Rust pending
Forgemaster
committed
To the Metal round 4: COBOL, ALGOL 60, Lisp, Autocode, Short Code + back-translation analysis
Forgemaster
committed
To the Metal round 5: COBOL (476 lines) + ALGOL 60 (570 lines) complete
Forgemaster
committed
To the Metal round 6: Non-von-Neumann architectures + forgotten paradigms
Forgemaster
committed
To the Metal round 7: wenyan (文言) — Deadband Framework in Classical Chinese
Forgemaster
committed
To the Metal round 8: GPU mega benchmark + Vedic (Sanskrit) deadband
Forgemaster
committed
To the Metal round 9: Mass Horizontal States — 7 GPU experiments
Forgemaster
committed
To the Metal round 10: wenyan 21/21 + Vedic ALL PASS — tested at the metal
Forgemaster
committed
To the Metal round 11: 25 novel deadband-native language features (DeepSeek)
Forgemaster
committed
To the Metal round 12: Vedic compiler 7 builtins + Rust crate 129K zero-drift + GPU 2.12× swarm speedup
Forgemaster
committed
To the Metal round 13: C AVX2 23.4x snap speedup + master comparison
Forgemaster
committed
To the Metal round 14: Fortran beats C 2.4x — auto-vectorized to AVX2 for free
Forgemaster
committed
docs: SuperInstance + Cocapn org READMEs updated
Forgemaster
committed

Commits on May 20, 2026

chore: add [workspace] table for crates.io publish
Forgemaster
committed
chore: add [workspace] table for crates.io publish
Forgemaster
committed
Add FLUX-SYNTHESIS.md — grand synthesis essay tying constraint theory ecosystem together
Forgemaster
committed
fix: all 3 test failures — BMA indexing, fib-spline threshold, hpdf variance
Forgemaster
committed

Commits on May 21, 2026

fix: add stdint.h for CUDA 11.5 compatibility
Forgemaster
committed
fix: remove non-existent workspace members, deduplicate imports
Forgemaster
committed

Commits on May 22, 2026

doc: deep synergy analysis — the one pattern that binds 1,681 repos
Forgemaster
committed
doc: unified proof (589 lines) + fleet protocol + core design
Forgemaster
committed
fix #20: forgemaster build fix
Forgemaster
committed
audit: FLEET-STRATEGIC-AUDIT.md — brutally honest strategic review of 1,681 repos
Fleet Auditor
committed
feat: Cocapn fleet metronome simulation
Fleet Auditor
committed
chore: preserve all working state
Fleet Auditor
committed
chore: commit working state
Fleet Auditor
committed
docs: streamline forgemaster into clean coordinator shell
Fleet Auditor
committed
refactor(python): deep code quality pass across all modules
OpenClaw
committed
fleet: coordination message to kimi1 — blockers cleared, task menu
OpenClaw
committed
fix: MIDI export, pyproject.toml, adapters package init
OpenClaw
committed

Commits on May 23, 2026

Polish repo: organize root, create docs/ structure, write professional README
OpenClaw
committed