Skip to content

Commands#

Trident CLI supports the following commands:

trident init#

Initializes Trident Workspace and generates a new Fuzz Test Template. Works with both Anchor and vanilla Solana programs.

Creates the following structure:

project-root
├── trident-tests
│   ├── fuzz_0 # particular fuzz test
│   │   ├── fuzz_accounts.rs # Account addresses storage
│   │   ├── test_fuzz.rs # Main fuzz test logic
│   │   └── types.rs # IDL-like generated types
│   ├── fuzz_1
│   ├── fuzz_X # possible multiple fuzz tests
│   ├── target
│   ├── Cargo.toml
│   └── Trident.toml # Configuration file located in trident-tests directory
└── ...

Options#

  • -f, --force - Force Trident initialization. Trident dependencies will be updated based on the version of Trident CLI.
  • -s, --skip-build - Skip building the program before initializing Trident.
  • -p, --program-name <FILE> - Specify the name of the program for which fuzz test will be generated (Anchor projects only).
  • -t, --test-name <NAME> - Name of the fuzz test to initialize.
  • --idl-path <FILE>... - Path(s) to IDL file(s). Multiple files can be specified separated by spaces. Required for vanilla Solana programs.

Anchor Projects#

# Default - auto-detects Anchor.toml and uses target/idl/
trident init

# With custom IDL paths
trident init --idl-path ./custom-idls/my_program.json

# Specific program only
trident init --program-name my_program

Vanilla Solana Projects#

For programs not written with Anchor, you must provide IDL file(s):

# Single program
trident init --skip-build --idl-path ./idl/my_program.json

# Multiple programs
trident init --idl-path ./idl/program1.json ./idl/program2.json

IDL Generation for Vanilla Solana

Vanilla Solana programs require IDL files to be generated using external tools or written manually following the Anchor IDL format. The build command uses cargo build-sbf instead of anchor build.

trident how#

Show the HowTo message about writing fuzz tests.

trident fuzz#

Run fuzz subcommands. With fuzz subcommands you can add new fuzz test template or you can run fuzz test on already initialized one.

Examples:

trident fuzz add
trident fuzz run fuzz_0
trident fuzz debug <FUZZ_TARGET> <SEED>
trident fuzz refresh fuzz_0

trident fuzz add#

Generate new Fuzz Test template. Can be executed from the project root or from within the trident-tests directory.

Options#

  • -p, --program-name <FILE> - Specify the name of the program for which the fuzz test will be generated (Anchor projects only).
  • -t, --test-name <NAME> - Name of the fuzz test to add.
  • -s, --skip-build - Skip building the program before adding new fuzz test.
  • --idl-path <FILE>... - Path(s) to IDL file(s). Multiple files can be specified separated by spaces. Required for vanilla Solana programs.

trident fuzz run <fuzz_target>#

Directory Note

Execute fuzz tests from the trident-tests directory.

Solana Logs

In case you want to see the logs of the fuzzed transactions, prefix the command with TRIDENT_LOG=1. 

TRIDENT_LOG=1 trident fuzz run <fuzz_target>

Runs the specified Fuzz Target using Trident's Manually Guided Fuzzing (e.g., fuzz_0).

Arguments#

  • <fuzz_target> - Name of the desired fuzz template to execute (for example fuzz_0).
  • [seed] - Master seed used for fuzzing, if not provided it will be generated randomly.

Options#

  • -e, --exit-code <MODE> - Exit with non-zero code on failures. Available modes:

    • all - Exit non-zero on any failure (program panics or invariant failures)
    • invariants - Exit non-zero only on fuzz test assertion/invariant failures

Without --exit-code, Trident does not fail the process on invariant failures or program panics.
Unexpected fuzz-test panics (for example unwrap() on None) are always treated as runtime errors and fail the run.

Examples:

# Run without exit code handling
trident fuzz run fuzz_0

# Exit non-zero on any failure
trident fuzz run fuzz_0 --exit-code all

# Exit non-zero only on invariant failures (useful in CI to catch assertion bugs)
trident fuzz run fuzz_0 -e invariants

trident fuzz debug <fuzz_target> <seed>#

Debug crashes by analyzing specific crash files using the provided seed.

Arguments#

  • <fuzz_target> - Name of the desired fuzz template to execute (for example fuzz_0).
  • <seed> - Master seed of the desired fuzz template to execute.

Debug Output

The debug output includes:

  1. Transaction logs
  2. Instruction data structures
  3. Panic/Crash information

trident fuzz refresh <fuzz_target>#

Refresh an existing fuzz test by regenerating the types file based on the current program state. Can be executed from the project root or from within the trident-tests directory.

Arguments#

  • <fuzz_target> - Name of the fuzz test to refresh (for example fuzz_0).

Options#

  • -p, --program-name <FILE> - Specify the name of the program for which the fuzz test will be refreshed (Anchor projects only).
  • -s, --skip-build - Skip building the program before refreshing the types file.
  • --idl-path <FILE>... - Path(s) to IDL file(s). Multiple files can be specified separated by spaces. Required for vanilla Solana programs.

trident clean#

Clean build target, additionally perform anchor clean.

trident server#

Start HTTP server to serve fuzzing dashboards. More info in the dashboard section.

Options#

  • -d, --directory <DIR> - Directory to monitor for dashboard files (default: .fuzz-artifacts).
  • -p, --port <PORT> - Port to run the server on (default: 8000).
  • --host <HOST> - Host to bind the server to (default: localhost).

trident compare#

Compare two regression JSON files and identify differing iteration seeds. More info in the regression section.

Arguments#

  • <FILE1> - Path to the first regression JSON file.
  • <FILE2> - Path to the second regression JSON file.