EXTENSIONS
Built by operatives — models, drivers, vaults, and reports, the parts that plug into Swamp.
Filter by what you need and pull what fits.
Bootstrap
Bootstrap a project for any Swamp-integrated agent with a reviewed architecture control plane and the official Swamp software-factory engine. Captures content-addressed accepted baselines, preserves existing project files, reserves one factory and a unique verification workflow per work item, and checks workspace isolation and source-digest provenance. Includes guided project design, personal working policy, factory-driving skills, and a readiness report that distinguishes configuration acceptance from successful verification. Local intake works without tracker credentials; selected tracker extensions supply remote reads.
Container Verification
Docker verification with structured read-only source mounts, cancellation protection, and retained evidence. Reuses the official container-image runner.
Software Factory
A fully generic, model-driven state machine for guiding agents through a
Software Factory Flow Metrics
Deterministic quality, reliability, flow, and ceremony metrics for a @swamp/software-factory work item — time-to-terminal, per-stage durations, entry counts and time-to-gate, dispatch attempts, failed/parked stage, human touches, deduplicated human decisions keyed by gate+stage+cycle+decision, approvals and rejections per gate, approval wait durations, stage visit/cycle counts, review/patch frequency, stage yield and park rates, time to verified draft, bounded-loop exhaustion, human cycle-limit overrides and the stages they unblocked, delivery mode, and terminal outcome — with a cross-run aggregate, every metric carrying a trust/availability label and a source pointer back to the journal, approval, or artifact record it came from, so unmeasurable values are reported as unavailable rather than as zero. Rendered statically from recorded run data with no LLM involved.
Software Factory Run Audit
Deterministic run reconstruction and invariant-violation audit of a @swamp/software-factory work item — stale SHAs across the artifact lineage, impossible states, unresolved critical/high blockers, and retained resources — rendered statically from recorded run data with no LLM involved, every flag traceable to the record it came from. Optional retained-worktree and external issue-tracker (Linear) correlation joins are pluggable, not required.
Eda Review Lenses
Five adversarial EDA design-review prompts ("lenses") for an independent
Go Review Lenses
Three adversarial Go code-review prompts ("lenses") for an independent
External Reviewer
External, independent adversarial review for swamp software factories.
Meta Factory
Deterministically assemble a @swamp/software-factory definition from the design
Kw Review Lenses
Three adversarial knowledge-work (prose/document) review prompts ("lenses")
Vale Review
Run the Vale prose linter as an independent, deterministic reviewer for a
Sdlc Review Lenses
Two language-agnostic adversarial review lenses for the SDLC artifacts a
Ts Review Lenses
Three adversarial TypeScript code-review prompts ("lenses") for an independent
Software Factory
A fully generic, model-driven state machine for guiding agents through a
Software Factory
A fully generic, model-driven state machine for guiding agents through a