Crowdsourced Project Mesh Economics
Contractors self-fund exploration tokens. Sponsors reward verified impact. SISAGA language is the constitution of crowdsourced work, not the entry barrier for every contractor.
Append-only project memory
Spark Log
The living record of SISAGA's naming, principles, visual direction, product judgments, collaboration rules, and implementation sparks.
Back to homepageCore rule
Never delete the memory.
Existing Spark Log content is preserved. If an older idea becomes incomplete, naive, or wrong, SISAGA agents add a newer dated refinement instead of rewriting history.
Spark entries are not formal specs. They are the high-signal raw material that later becomes docs, roadmap, runtime behavior, and implementation guidance.
Recent sparks
Latest project signals
Agent Engineering Lease System
SISAGA should not build an Agent group chat. It should turn human intent into a project contract, split work into packages, grant limited capability leases, require evidence, and record contribution.
Engineer the Ellipsis
SISAGA should not own the place where people state goals. It should engineer what happens after the goal: contract, permission, coordination, evidence, reuse, contribution, and memory.
Memory and Skill Loops Need Evidence Boundaries
OpenClaw, Hermes Agent, and OpenHuman create different protocol pressure: task evidence, skill provenance, and private memory context should enter SISAGA through typed evidence boundaries.
Do Not Copy Goal Surfaces
SISAGA should not copy Codex goal mode, Claude outcome specs, Cursor work items, Gemini interactions, or OpenClaw task channels. It should standardize the I2P contracts, context artifacts, evidence, profiles, skills, and contribution records underneath them.
SI-Dev-RI Strategic Guardrail
SI-Dev-RI exists to keep SISAGA's development trajectory ahead of the frontier and strategically correct, so the project is not displaced during its own development cycle by homogeneous new agent products.
Deep Thought Is Not a Fact Source
Deep Thought records strategic judgment, while SI-Dev-RI records source-linked facts. GPT Pro reasoning can shape priorities, but official sources, implementation, tests, benchmarks, gates, and evidence remain the proof layer.
SI-Dev-RI as Development Reference Intelligence
SI-Dev-RI is the proposed SISAGA Development Reference Information skill: not a news feed, but a source-linked radar that turns agent product updates into next-version engineering decisions.
Agent Release Notes as Primary Radar Signal
Version updates from major agent products are first-class strategic intelligence. SISAGA should track which capabilities become productized, then decide what belongs in its next contract, runtime, artifact, gate, or benchmark.
Deep Thought Channel Applies Radar to Next Version
Each completed Frontier Radar round should create a bilingual Deep Thought record that connects AI frontier signals and GPT Pro strategic reasoning to the next SISAGA development version.
Frontier Radar With Pro Reasoning Loop
SISAGA should evolve through a repeatable loop: fresh AI frontier signals, current project state, GPT Pro strategic dialogue, impact judgment, then implementation and verification.
Protocol Surface Governance
SISAGA's protocol surface must be registerable, normalizable, migratable, and gateable. Without that, a growing ecosystem of runners, adapters, and skills becomes unauditable fragmentation.
Language Implementation and Bootstrap Strategy
SISAGA separates protocol from implementation. It should bootstrap with Python, move its hard runtime and policy kernel toward Rust, use TypeScript for developer surfaces, and let SISAGA contracts govern the project's own evolution before selected components become self-hosted.
Agent-Native Development Language and Performance Contracts
SISAGA is both an Agent-native development language and a portable work contract. Its long-term performance route is benchmark-driven synthesis: performance contracts, many generated implementations, strong backends, profiling, and reproducible evidence.
Project Version Separates from Spark Version
SISAGA introduced an independent project version line. Spark versions remain the append-only idea ledger, while roadmap labels such as v0.2 stay future capability milestones.
Rollback Becomes a Standard Artifact
Every sisa run now writes rollback.json, recording whether rollback is required, why, the declared rollback block, and actionable recovery steps for future SAGA execution.
Trace Replay Makes Evidence Reviewable
sisa replay now reads run artifacts and prints a compact summary of task identity, policy status, verification result, event count, policy errors, and command outcomes.
Scope Starts Enforcing Local Command Paths
The V0.0.0.1 preview policy guard now checks local paths referenced by verification commands against scope.include and scope.exclude, blocking explicit command-path drift before execution.
Human Approval Becomes a Runtime Gate
Manifests with approval.required now block before verification commands run unless the operator explicitly passes --approve, making human review an enforceable contract boundary.
Policy Kernel Starts Enforcing Denied Effects
The V0.0.0.1 preview runner now evaluates command effects before execution and writes policy.json, turning denied effects such as network, deployment, secret reads, and token transfer into enforceable runtime checks.
SAGA as Governed Execution Flow
SAGA is reserved as SISAGA Agentic Governance Automaton: the governed execution flow that runs a SICO contract through planning, verification, compensation, rollback, evidence, and contribution finalization.
SICO as Contract Object Layer
SICO is reserved as SISAGA Contract Object: a precise name for the contract layer, file object, or protocol submodule without replacing SISAGA as the main language and civilization protocol.
Dogfooding Gate and Self-Benefit Threshold
SISAGA should switch its own development to contract-driven work only when internal tasks become clearer, safer, easier to review, and no slower than ordinary development.
Benchmark Claims and Measurement Discipline
Productivity and quality claims must distinguish measured results, hypotheses, and acceptance thresholds. Large multipliers only become public claims after controlled A/B benchmarks.
Public Copy Governance
Public website copy now uses external product and protocol language: precise claims, professional audience labels, and no private discussion phrasing.
Architecture Channel and Ecosystem-First Post-1.0
SISAGA's strongest claim is software-production leverage, not raw performance supremacy. The site now gets a dedicated technical architecture channel, and the post-1.0 path turns toward global ecosystem building.
Experience Comparison and Visual Roadmap Revision
Orange-red becomes the dominant gradient language, violet stays a small accent, and the experience section compares SISAGA and non-SISAGA workflows directly.
Collision Map and SISAGA 1.0 Vision
SISAGA's durable distinction is the portable work contract layer: intent, permission, Agent IR, trace, evidence, and contribution record across languages, agents, harnesses, tools, and teams.
Implementation Language Is Not Protocol Language
Python is only the current bootstrap language for the V0.0.0.1 preview runner. SISAGA itself is the portable task contract, Agent IR, evidence, trace, permission, and contribution layer.
Selective Spark Updates
The Spark Log is a high-signal idea ledger, not a development log. Only durable inspiration, rules, strategy, architecture, visual direction, and governance ideas should be recorded.
Whitepaper Protocol Palette
The website palette shifts toward a premium protocol whitepaper: warm white, black type, orange heat, violet intelligence, and soft pink-lavender washes.
Spark Page and Version Format
Future spark versions use V0.0.0.X. The homepage Spark area opens this bilingual page. Unsupported future languages fall back to English.
Multilingual Coverage Is Product Work
Language switching must cover the actual collaboration surface, while technical names, commands, files, and protocol names may stay canonical.
Pretext Spatial Typography
The homepage should use restrained 3D spatial typography: floating text planes, grid depth, telemetry, and readable whitepaper content.
Motion and Multilingual Surface
The website should feel cinematic and alive, using timeline-driven motion while staying readable and globally collaborative.
Whitepaper Homepage Mandate
The homepage is both public window and collaboration cockpit: mission, project state, evidence, roadmap, and sparks near the top.