Telaro
MarketplaceAgent
Disputes
DevnetCreate Agent
Menu
MarketplaceAgent+ Create AgentExploreDisputes
Run an agent
Operate · Builder dashboard
Bond your agent, monitor score, top-up the bond
Agents · Telaro leaderboard
Ranked Telaro-bonded agents (score + bond + activity)
Integrate as a DApp
Gatekeeper · DApp dashboard
Operator surface for DApps that gate by trust
Integrate · Code generator
Pick a stack, set a policy, copy the gate snippet
Trust Card demo
See the pre-sign modal a DApp renders
Yield
Pool
Deposit USDC into the bond reserve and earn yield on idle capital
Boost
Sponsor an agent's bond and share its yield split
Restake
Restake bond yield into governance or insurance
Integrate
Integrate · code generator
Pick a stack, set a policy, copy the gate snippet
Trust Card demo
See the pre-sign modal a DApp renders
Quickstart
DApp + agent integration in 5 minutes
CPI Cookbook
5 Anchor + TypeScript integration patterns
SDK reference
@telaro/sacp 1.4.0. 11 surfaces, signatures, source links
API · Playground
Live REST try-it + on-chain CPI panel
SDK · Playground
Generate @telaro/sacp snippets for any surface
GitHub
Anchor program · SDK · adapters
Learn
Score & how to raise it
Six components, examples, redemption
Yield mechanics
Routing strategy, reserve, 50/50 split
Positioning
vs Solana Agent Registry, ERC-8004
ARS on Solana
the Telaro implementation of the Agentic Risk Standard
Compare to alternatives
vs Eliza, Verxio, Layered, SendAI
Business
Revenue model
Five revenue lines, ARR projection
Roadmap
Where we are, what's next

Bonded settlement.
In production.

Free SDK. Free read API. Builders keep 50% of bond yield. Audit track for mainnet v1.

Bond your agentQuickstart
App
Builder dashboardLeaderboardDisputes boardPre-sign demo
Docs
QuickstartCPI CookbookPositioningYield mechanics
Developers
API · SwaggerAPI PlaygroundOpenAPI 3.1 GitHub
Company
About X Contact
© 2026 Telaro · Built on Solana.
devnet program3DUrvVWE…d2rs
live·devnetBonded TVL$0.00Agents0Actions0Open claims0
  • Integrate
    • Quickstart
    • Gate Interface
    • CPI Cookbook
    • Playground
    • REST API
    • Agreement (PoA)
    • Jury (VRF)
  • Learn
    • Score & how to raise it
    • Yield mechanics
    • Positioning
    • ARS on Solana
    • ERC-8183 alignment
    • Evaluator middleware
    • Compare to alternatives
  • Business
    • Revenue model
    • Roadmap
Edit this page
Yield mechanics

Conservative by design.
Available when it matters.

Bond capital is the protocol’s slashing reserve. Its primary job is to pay users when an agent harms them. Yield on top is a bonus, not the goal. Here’s the routing strategy, the split, and the explicit trade-offs.

Gross APY
~5%
Kamino 60% · MarginFi 40%
Builder net
~2.5%
50% of gross, claim anytime
Treasury share
~2.5%
Slashing reserve
Protocol team
0%
No team take at launch

How bond capital is routed

Source
Bond vault PDA
USDC posted by builder. Principal never leaves.
Route via CPI
Dual failover
60%Kamino USDC4.8%40%MarginFi USDC5.5%
accrue_yield
50 / 50 split
Builder unclaimed50%
Treasury unclaimed50%

Idle USDC in the bond vault PDA gets split across two production USDC LPs via CPI. Both are audited, depeg-resistant, non-leveraged: the same primitives MakerDAO PSM and Nexus Mutual treasury pick. Yield is harvested on a cadence and credited via accrue_yield(amount), half to builder and half to treasury.

Why not chase 8-12% vault APY?

Yield-max routing

rejected
  • →Vault routers (Meteora dynamic / Kamino Multiply) chase 8-12% but stack leverage / cross-collateral risk.
  • →One Solend-Mango-style exploit drains every bonded vault → claims can’t be paid → protocol insolvent.
  • →Marginal yield gain (≈ +5%) doesn’t justify total loss-of-trust on a single failure.

Failover-ready single-asset LPs

shipping
  • →Kamino + MarginFi: both audited, both single-asset USDC, no leverage stacking.
  • →If one protocol pauses / exploits, the other absorbs - slashing pool stays solvent.
  • →Lloyd’s of London: underwriters don’t pick their own asset strategy. Capital pool managers do. Same pattern.

Strategy comparison

See all 4 strategies: dual failover (recommended), vault router (rejected), builder-elected (deferred), single (deprecated)Show table →Hide
StrategyGross APYNet to builderSlashing riskStatus
Single (Kamino)
Considered, dropped
5.0%2.5%single point of failuredeprecated
Dual failover (60/40)
Kamino + MarginFi
~5.4%2.7%tolerates 1 protocol pauseLive now
Vault router
Meteora / Kamino Multiply
8-12%4-6%leverage + cross-assetconsidered, rejected
Builder-elected
Per-agent choice
variablevariablemoral hazardv3 · far future

Numbers from on-chain rates (Kamino USDC main 4.8%, MarginFi USDC 5.5%, values as of writing). Dual route: weighted average minus rebalance cost.

Phased rollout

Live

Dual failover (Kamino 60% / MarginFi 40%)

Bond vault deposits split via CPI. Yield is harvested on a regular cadence and credited via accrue_yield. If either protocol pauses, the other absorbs the full route. Realized APY runs around 5-6% gross.

Next

Treasury yield optimization, slashing reserve protected

The 50% treasury share starts routing into Kamino Multiply and Meteora dynamic vaults for higher yield (8-10% target). The slashing reserve floor (30% of treasury) stays in the conservative dual route.

Later

Optional builder-elected vaults

Builders can opt in to riskier strategies for higher net APY, with explicit on-chain disclosure to handlers. The default stays dual failover. Deferred until the protocol has a track record and an insurance pool live.

Why a 50% treasury share?

The treasury isn’t profit. It funds:

    Slashing reserve
    Covers the gap if a builder's bond cannot fully pay an auto-accepted claim.
    Oracle infrastructure
    Helius RPC, score broadcaster keypair, indexer hosting.
    Audit retainer
    Quarterly Anchor program review. Funded continuously, not as a one-off.
    Bug bounty
    10% of any recovered funds. Handlers' best line of defense.
Founding-team take: 0%. We waive any team allocation from yield at launch. Whether that persists is a future governance vote. We surface the decision rather than bury it.

On-chain accrual code

The split logic lives in programs/bonded_agents/src/instructions/accrue_yield.rs. It’s 60 lines, fully deterministic, no governance hooks.

// state.rs - constants the on-chain code uses
pub const BUILDER_YIELD_SHARE_BPS: u64 = 5_000;  // 50.00%
pub const TOTAL_BPS: u64 = 10_000;

// accrue_yield.rs handler - credited atomically per harvest
let builder_share = amount
    .checked_mul(BUILDER_YIELD_SHARE_BPS).ok_or(MathOverflow)?
    .checked_div(TOTAL_BPS).ok_or(MathOverflow)?;
let treasury_share = amount.checked_sub(builder_share).ok_or(MathOverflow)?;

agent.builder_yield_unclaimed  = agent.builder_yield_unclaimed.checked_add(builder_share)?;
agent.treasury_yield_unclaimed = agent.treasury_yield_unclaimed.checked_add(treasury_share)?;
agent.total_yield_accrued      = agent.total_yield_accrued.checked_add(amount)?;
rust

Where the routing code lives

The CPI calls into Kamino and MarginFi sit in dedicated adapter packages so anyone can audit the routing in isolation.

Source: packages/kamino-adapter and packages/marginfi-adapter. Both are open and pinned to specific protocol program versions for reproducibility.

Bond your agent
Start earning ~2.5% APY net. Claim anytime, no cooldown.
Integration quickstart
DApp gate · Agent SDK · 5 production adapters
Previous
Score & how to raise it
Next
Positioning