Struct SequencePlan 

pub struct SequencePlan {
    pub sequences: Vec<Sequence>,
    pub master_seed: u64,
    pub timeout: Duration,
    pub transport_name: String,
    pub severity: SeverityConfig,
    pub fuzz_corpus: Option<FuzzCorpus>,
    pub mutate_ratio: f64,
}

Plan for executing a batch of Sequences.

Fields

sequences: Vec<Sequence>

Sequences to run, in declaration order.

master_seed: u64

Master seed; per-sequence seeds derive from master_seed + sequence_name deterministically.

timeout: Duration

Per-step call timeout.

transport_name: String

Transport label for ReproInfo.

severity: SeverityConfig

[severity] overrides from wallfacer.toml.

fuzz_corpus: Option<FuzzCorpus>

Phase V — optional persistent corpus, shared with the v0.6 fuzzer. When set, sequence steps mutate from the corpus mutate_ratio fraction of the time and save inputs that trigger findings or produce a previously-unseen response fingerprint per step. Cross-pollinates with wallfacer fuzz --corpus-feedback: fuzz-discovered “interesting ids” can seed sequence steps that call the same tool.

mutate_ratio: f64

Phase V — fraction of sequence-step inputs that mutate from the corpus instead of using the YAML literal verbatim. Range 0.0..=1.0. Default 0.9 matches the fuzz default. Ignored when Self::fuzz_corpus is None or the per-tool corpus is empty.

Implementations

impl SequencePlan

pub async fn execute<C: McpExec + ?Sized>( self, client: &mut C, corpus: &Corpus, reporter: &mut dyn Reporter, ) -> Result<SequenceReport>

Drives the sequence loop. Returns once every sequence has either passed, produced a finding, or been skipped for a missing tool.

Lifecycle events (on_run_start / on_run_end) are not emitted: callers compose this plan with the property plan and run them through a single reporter instance, so wrapping each sub-run with its own start/end would split the JSON envelope and confuse downstream consumers. The reporter sees a clean stream of on_finding / on_skipped calls with the sequence findings interleaved with the single-tool findings.