Zycord

Audit date · 2026-10-17

This is a point-in-time review of code quality and tokenomics at the commit above, not a safety certification. Projects can change after this date. Check for newer releases or a newer audit before relying on these scores.

RandomX Go L1; no premine, height-based issuance, and a keyless, time-limited treasury.

🪙 Zycord (ZCD) — Code & Tokenomics Audit

Repo: gitlab.com/zycord-group/zycord-node at commit a580df8ccc081a678fd6bee2cf55cbdba48830c6
Date: 2026-10-17 Scope: recon, code review (money and consensus path read line by line), tokenomics, write-up. All phases ran. Hashrate and profitability were not measured — they are out of scope for this scale.

🎯 Assessment: FAVORABLE (reviewed code and tokenomics only; not a safety certification)

Zycord is a hand-written Go layer-1 with a RandomX v2 CPU subsidy and an unusually strict codebase. Issuance is a pure function of block height: every node recomputes the subsidy from the height alone, and no header carries an amount. There is no premine, no mint path to the native coin, and the founder share is 3% and time-boxed.

This scores code and tokenomics only. It says nothing about whether Zycord survives, whether anyone will trade it, or whether it outlives its author. The chain is about ten days old at audit and the author is anonymous. Those are survival questions, not trust problems, and they do not move the number.

🔑 At a glance

🔗 Links & sources

🧮 Risk score inputs

code deductions, one at a time:
  0.5
tokenomics deductions, one at a time:
  0.0

code_score       = 10 - 0.5                            = 9.5
tokenomics_score = 10 - 0                              = 10.0
risk_score       = min(code, tokenomics)               = 9.5

Instant-AVOID override fired: none. All five checked —

1. No key, role or admin path mints native supply or edits the schedule outside consensus. deriveMint refuses a MINT unless the asset carries version byte 0x03, so the all-zero native asset cannot be minted (core/validity/derive.go:216-224, core/types/slots.go:10-11).
2. No code sends a seed, key or wallet contents anywhere; the phrase path says so outright (wallet/phrase.go:23-25).
3. No backdoor, killswitch or remote-code path found in node/, core/, wallet/.
4. Not a fork presented as original work — RandomX is an external algorithm behind a build tag and is declared in the consensus root (spec/params.json:56).
5. No supply or privacy claim that the code contradicts — see the tokenomics table.

🔒 Code

Commit audited: a580df8

• ✅ The subsidy is never read from the block. Emission(height) is a pure function of height (core/params/params.go:256-266) and the split is recomputed per block at core/params/founder.go:58-70, called from core/fold/fold.go:402. A miner cannot declare a reward. The header carries only a payout address (core/types/block.go:60), validated as a user address at core/fold/blockrules.go:662-670 (B11).
• ✅ Coinbase rolls through a 240-slot maturity ring and burns the reward if the payee is spent (core/fold/fold.go:1395-1428, F12). Producer, founder and treasury credits are applied inside one journal, with undo on overflow (core/fold/fold.go:419-441).
• ✅ Height continuity is enforced where ancestry is held: node/chain/forkchoice.go:192,201 refuse a branch block that is not parent height + 1. This is the check that makes "issuance is a function of height" safe — a header cannot claim height 5,000,000 on top of height 3 and collect a mature subsidy.
• ✅ Proof of work is checked on every path: pow.CheckWork recomputes the RandomX digest (core/pow/pow.go:57-73), the cheap half rejects a target above max_target (core/pow/pow.go:95-105, B38), and the declared seed epoch is pinned to the height (core/fold/blockrules.go:651-657, B0b).
• ✅ Protocol cells are unreachable to user traffic twice over: V7 refuses any certificate write under address version 0x00 (core/validity/validity.go:618-624) and F13 refuses it again at apply (core/fold/fold.go:1591-1597). Debits elsewhere require a signature-bearing user address (core/crypto/crypto.go:164-166, used at core/validity/intent.go:282).
• ⚠️ LOW — PoWHash is carried in the header and compared rather than treated as secret. CheckCommitment runs where no parent is known and evaluates no work function (core/pow/pow.go:95-105, B38), so a header with an easy-looking target passes the free gate and then costs one RandomX hash to reject at core/pow/pow.go:69-72. Reachable on the gossip and sync paths (node/sync/sync.go:394, node/verify/work.go:42). Consensus is not broken — the digest is still verified against the header input. Fix: rate-limit work evaluations per peer.
• ✅ Wallet keys are encrypted at rest: Argon2id + AES-256-GCM with a validated parameter ceiling (wallet/keyfile.go:23-62,271-272). The BIP39 PBKDF2 step is deliberately skipped and the reason is written down (wallet/phrase.go:23-25).
• ✅ Node RPC binds loopback only, 127.0.0.1:9420, and rejects a non-loopback Host (node/rpc/rpc.go:54,270-281). No telemetry or analytics code found in node/, wallet/, cmd/.
• ✅ Dependency surface is three golang.org/x modules, confined to wallet/ and cmd/ by make check-imports; core/ is standard library only (go.mod:1-14). go.sum is present. The storage engine is written in-tree rather than imported. No unpinned-dependency deduction.
• ✅ unsafe.Pointer appears only in Windows syscall shims and a PTY test (internal/osfile/osfile_windows.go, node/storage/lock_windows.go, cmd/zycordd/underwrite_pty_linux_test.go). Nothing on a consensus path. No deduction.
• ✅ Tests cover the money path heavily: core/validity/emission_bound_test.go cross-checks CumulativeEmission against a height-by-height sum of Emission (lines 113, 132-155); core/fold/theorem_test.go:298,340 pin that the supply cap, not a per-cell ceiling, stops a mint; core/cevm/protocolcellguard_test.go:57 sweeps opcodes for guards on protocol cells including the treasury cell. No deduction.
• ✅ CI exists and builds the release artifact: .gitlab-ci.yml is a tag-triggered release stage that requires the signing key and deploy credentials to be mounted and refuses a tag that is not on the commit. No deduction.

📊 Tokenomics

🕵️ Privacy

• ✅ RPC binds loopback only, 127.0.0.1:9420, and refuses a non-loopback Host header as CSRF protection (node/rpc/rpc.go:54,270-281). Remote access is documented as an SSH tunnel or an authenticating reverse proxy (node/rpc/rpc.go:399).
• ✅ No telemetry or analytics code in node/, wallet/, cmd/, internal/. No forced outbound beacon found on the node or wallet paths.
• ✅ The update system requires a detached signature and treats its absence as fatal (update/carryforward.go:11-14); the release key is committed publicly and the CI job compares rather than copies it (.gitlab-ci.yml).
• ✅ Key custody is local. Key files are Argon2id + AES-256-GCM, with the passphrase as the encryption (wallet/keyfile.go:23-62,271-272). Nothing leaves the machine.
• ⚠️ The BIP39 PBKDF2 seed step is deliberately not performed anywhere (wallet/phrase.go:23-25). The paper phrase is not a second factor — the key-file passphrase is the only encryption. Pick a long passphrase.
• ⚠️ Stratum mining links your IP to the payout address like any pool (node/stratum/conn.go:424). Solo mining avoids it.
• 🔍 Exit privacy is unconfirmed for this audit — it depends on where the coin is traded, which is out of scope.

🚩 Red flags

• ⚠️ LOW — the header declares its own PoWHash, so the free pre-execution work gate passes an easy-looking header and a full hash is needed to reject it (core/pow/pow.go:57-73). A per-peer resource shape, not a consensus break.
• ⚠️ RandomX key schedule departs from upstream: the key comes from height plus chain id, not a key block hash (spec/params.json:66). Every future key is known from genesis. The author states the cost himself: long-horizon offline specialisation. The PoS gate at height 1,576,800 bounds the exposure.
• ⚠️ Single anonymous author (Simstoshi) and a chain about ten days old at audit — genesis_time 1791331200 is 2026-10-07 00:00 UTC. Ordinary abandonment risk, not a trust problem.
• ⚠️ The launch was relaunched: commit b4303e1 (2026-10-03) pins mainnet and testnet chain ids live after "every relaunch parameter frozen", and the params file records an earlier h1_bond of 40,320 that was superseded before launch. The cancelled instance's chain id stays spent in spec/chain-ids.json.

✅ Green flags

• ✅ Issuance is a pure function of height and every node recomputes it: core/params/params.go:256-266 read at core/fold/fold.go:402, and re-applied to user claims at core/validity/validity.go:755.
• ✅ No premine, and the refusal is written into both the emission function and the genesis shape rule (core/params/params.go:258-260, core/fold/blockrules.go:643).
• ✅ The treasury is a keyless protocol cell — 3%, then 6%, of the subsidy accrues there and no transaction can move it (core/validity/validity.go:618-624, core/crypto/crypto.go:165-167).
• ✅ Anti-wrap thinking in the parameter validator, core/params/founder.go:75-95.
• ✅ Consensus test volume far above the norm for a small PoW chain: core/fold is about 27,000 lines including property_test.go, griefing_era1_test.go, cert_budget_test.go, theorem_test.go; golden vectors in spec/ plus a 3,253-line vector generator.
• ✅ Supply-chain discipline: three third-party modules total, storage engine in-tree, allowlist enforced by make check-imports (go.mod:1-14). The reproducible-build pin lives in the Makefile and a test fails if go.mod, the Makefile and the Dockerfile disagree.
• ✅ Documentation states its own costs, including the 907,917.44 ZCD the slow start withholds and never repays (spec/params.json:64). Reproduced independently below.
• ✅ A parameter file cannot be loaded unvalidated: params.Parse sets DisallowUnknownFields() then calls Validate() before returning (core/params/params.go:735-742), and the comment there states why Validate alone is not the gate. A typo'd founder field fails to load rather than defaulting.
• ✅ Checkpoints cannot mint. They are client history pins, not consensus: spec/checkpoints.json carries a _not_consensus key saying nothing in it enters params.ConsensusRoot(), mainnet's checkpoints array is empty and min_chain_work is 0, and the set exposes only refusal-style checks — Contradicts, MeetsFloor, ValidateAgainstTip (node/checkpoints/checkpoints.go:94,101,122) — and no balance-write path.
• ✅ Emission reproduced independently from spec/params.json alone (script /tmp/audit/zycord/emcheck.py, replaying buildEmissionTable at core/params/params.go:369-390): block 1 pays 0.00023623 ZCD and the slow start withholds 907,917.44 ZCD — both exact matches to the doc comment at spec/params.json:63-64. Rate at block 86,400 reproduces as 20.41043 ZCD against the doc's "20.41 ZCD".

📋 Context (does not affect the score)

❓ What I could not check

• 🔍 On-chain supply and the real block-1 payout. Still open: https://explorer.zycord.com/api/block/1 returns 404 page not found, so no public endpoint answers. Close it by syncing a node and comparing CumulativeEmission(h) with the sum of balance cells at height h. The code proves the schedule; it does not yet prove the live chain followed it.
• 🔍 One number I could not reproduce exactly: my independent replay of buildEmissionTable gives 4,377 table epochs and a tail handover at height 12,605,760, while spec/params.json:63 states epoch 4376 / height 12,602,880 — a two-epoch (one-epoch-index) difference at the tail boundary, which also shifts my pre-tail total to 62,745,788.87 against the documented 62,744,838.47. My script is naive about the final partial epoch, so this is recorded as my gap, not a repo defect; the repo asserts these values in its own tests. Close it by running go test ./core/params and reading the emission table tests rather than re-deriving them.
• 🔍 Founder address history beyond the code path. Close it by reading NativeBalanceSlot(0x02f29dc4...abe9b11) on a synced node and checking every credit came from a block subsidy.
• 🔍 Contributor count and per-commit authorship. Close it with a full clone plus GitLab API access.
• 🔍 Absolute hashrate and 51% cost. Out of scope for the score, but it stays the practical risk for a miner until measured from a synced node.
• 🔍 That the treasury cell has only ever accrued on the live chain. The code path is closed by V7 + F13 + IsUserAddress, so this is a belt-and-braces check.

🧭 Bottom line

This is one of the better-engineered small PoW chains reviewed here. The subsidy is a pure function of height, every node recomputes it, height continuity is enforced before the subsidy is read, and the treasury share has no spend path at all — it accrues into a protocol cell that no certificate can debit.

Tokenomics scores 10.0 because there is nothing to deduct: no premine in code, no native mint path, a 3% founder share that ends at height 3,153,600, and a deliberate perpetual tail of 0.33 ZCD per block from height 12,602,880. There is no hard cap and the tail continues forever — the project states that as design, and issuance being a pure function of height means an absent MAX_SUPPLY constant is not a defect.

Code scores 9.5: one LOW finding about the cheap work gate accepting a header-declared PoWHash. That is resource exhaustion per peer, not a consensus break. To reach 10.0 the node needs per-peer rate limits on work evaluations.

What would move it down a band: a parameter change that raises founder_share.bps or moves end_height without the chain-id ledger refusing it, or evidence of specialised RandomX hardware exploiting the fixed key schedule. Watch the PoS gate at height 1,576,800 and the founder end at 3,153,600.

Citations

core/pow/pow.go:57-73

core/pow/pow.go:95-105

node/verify/work.go:42

spec/params.json:35-38

core/params/params.go:269-276

core/params/params.go:256-296

core/params/founder.go:58-70

core/validity/validity.go:618-624

core/validity/derive.go:216-224

core/types/slots.go:10-11

Every citation on this page points at a file that was opened during the audit. Nothing here is a guess.

Back to Projects