Meaning System.io
Meaning must move. Maenen measures its integrity.
If meaning does not remain in motion and is compressed for convenience, such as icons, symbols, flags and dogma, it risks becoming rigid, stale and brittle. Yet the transportation of meaning risks divergence, dilution and drift.
Just as music and dance are amorphous movement, so must meaning flow. Subjective and shifting in value according to individual perspective, it can nevertheless be objectively measured at points of transformation.
MeaningSystem.io offers AI workflows Maenen's alignment, integrity and trust APIs as preset, evidenced route configurations for your agents.
— turning the philosophy of meaning-in-motion into practical checks at the transition points and boundaries where meaning transforms and transacts
— with preflight preparation, route evidence, drift detection and token-spend expectancy on every run.
MeaningSystem.io is not an agent harness.
Your agent does your work. You gain evidence it stayed true.
Secure browser account access is implemented locally. Production email activation is still pending, so this public panel is not operational.
Deterministic Core
Signed Receipts
No Ordinary Raw-Content Logging
Maenen is meaning in motion across representation, interpretation and transformation.
It does not solve the 'map/territory problem' by throwing away the map...
...it preserves and reveals the relationship between map and territory.
MeaningSystem provides an independent layer for checking what was requested, what changed, and what the evidence supports.
With lesser models, Maenen catches more errors. With frontier models, Maenen governs more consequential transitions.
So whilst more capable agents can perform more complex work.
MeaningSystem provides an independent layer for checking what was requested, what changed, and what the evidence supports.
Whichever the model category you use — never let the producing model be the sole authority on whether its own work conforms.
→ movement requires changing representations → representations are maps → preserve their relationship to what they represent.
AI pipelines lose original intent through compression, retrieval, vectorisation and heuristic transformation. Multi-step agent work ends in a black-box “Task complete” — with no answer to what actually happened, which tools were used, whether your intent survived, or what each step cost.
Worse, the same agent that produced the output is usually the one judging it. That is not verification. That is a witness testifying on its own behalf.
MeaningSystem answers with an external semantic conscience — not moralising, but measuring. Deterministic checkpoints watch the transitions, keep the evidence, and return or reference evidence for the work performed.
Submitted raw content is not intentionally placed in ordinary application logs; account, usage, receipt and evidence records are retained where the service contract requires them.
→scope→conditions→evidence→authority→qualifiers→acceptance criteria
Everything above decays in transit. MeaningSystem keeps it explicit, inspectable, testable.
One bounded route in. Inspectable assessments and linked receipts out.
Preset routes
Established, versioned compositions of the Maenen APIs — so a workflow never has to guess which checks to run, in which order, carrying which evidence.
Preflight preparation
Before spend, see the acceptance criteria and verification checklist for the route. Know what will be checked — and what a pass actually means.
Token-spend expectancy
Every route carries a declared MC bundle and per-stage attribution. Estimated before launch, evidenced on the receipt after.
Signed receipts
Route stages return or reference contract-specific evidence of the work performed — which may include timing, node path, spend and proof hashes. Compare runs. Inspect pathways.
Drift & integrity trace
A read-only trace of every transformation: which stages passed, cautioned, were rejected or skipped — and what remains unverifiable.
Capsules
Save or export a completed run as a verifiable capsule. Inspect its provenance and proof structure without restoring or rerunning the execution.
Current capability: /run · save · export · retain · verify
Choose the checkpoint. The route knows the instruments.
Select a route to see the question it answers and its bundle price.
Align Brief
Turns a goal into an inspectable intent scaffold: what is intended, required, and preserved — before any work begins. The preflight conscience for a new task.
- scope · conditions · acceptance criteria
- authority and qualifiers made explicit
- anchor for later drift checks
Secure Meaning Run
The full governed pathway: intent scaffold, integrity and drift checks across the transformation, cost accounting, and a signed proof surface. The default route when an agent submits its whole task and asks — evaluate it.
- preflight acceptance + verification checklist
- stage receipts with proof hashes
- trace / DAG + capsule export
Compare Meaning
Semantic difference between two sources — not a surface diff. Flags breaks in meaning between input and output, document and revision, promise and delivery.
- meaning-level divergence, not text-level
- anchored to the stated intent
- signed comparison receipt
Evidence Check
Are the claims grounded and the references valid? Checks supplied evidence against the statements that lean on it — within the declared evidence boundary.
- claim ↔ evidence binding
- unsupported additions flagged
- honest “unverifiable” where honest
Cost / Value Review
Was the spend proportionate? Assesses token cost ≠ efficiency step by step — an inexpensive check could prevent costly time and token waste.
- per-stage MC attribution
- value coherence assessment
- waste candidates identified
Agent Handoff Audit
When one agent delegates to another: did meaning, constraints, authority and responsibility survive the handoff? The checkpoint for coordinated swarms.
- delegation fidelity trace
- authority & constraint survival
- responsibility boundary record
→ Plus four A2A Trust checkpoints: claim assurance, artifact review, exchange review, transaction preflight.
Keep your agent stack.
Add the missing Trust layer.
A valid signature answers only part of the picture. Cryptographic integrity does not by itself establish provenance, applicability, authority, readiness — or a legitimate outcome.
MeaningSystem keeps those distinctions visible: four bounded trust checkpoints for claims, artifacts, exchanges and proposed transactions. Read-only. Non-authorising. Inspectable on return.
“Task complete” is a claim.
A receipt is evidence.
A completed MeaningSystem route returns or references contract-defined evidence: what was assessed, what changed, what it cost and whether the transformation remained faithful to the stated intent. Raw submitted content is not intentionally placed in ordinary application logs; account, usage, receipt and evidence records may persist.
route: secure_meaning_run_v1
status : VERIFIED — intent preserved
nodes : 6 / 6
drift 0.02 · within tolerance
spend : 200 MC (bundle) · $0.20
input_hash : 9f2c…e41a
output_hash : 77b0…03cd
root_hash : c4d9…51ff ✓ signed
capsule : ready · export available ──────────────────────────────
→meaning should survive processing — not require rediscovery downstream.
An inexpensive fuse for every downstream operation.
Maenen Credits measure deterministic work — not LLM tokens. No generative black box in the loop.
Hold, rejected, failed or missing-evidence trust outcomes cost nothing. Only completed positive conclusions are chargeable.
Retain and compare proven evidence. Small checks now can prevent large token waste later; capsules do not replay execution.
Meaning must move...
Move with Proof of Meaning.