This is the entry point for durable knowledge in the Fallow repository. Use the task context map to load only what a task needs.
Public user documentation is authored in
fallow-docs. The files below are
for contributors and maintainers of the open-source codebase.
- Development documentation: routing for shared architecture and workflow references.
- Knowledge architecture: ownership, trust boundaries, synchronization, and promotion policy.
- AI tooling: canonical skills and generated adapters for Codex and Claude.
- Task context map: ordered reads by task.
- Repository map: crates, pipeline, and subsystem hotspots.
- Quality gates: local verification and hook parity.
- Review routing: review lenses by changed path.
- Release procedure: maintainer publication steps and the invariants the policy tests assert
- Release security: credential boundaries and publication workflow ownership.
- Architecture invariants: crate boundaries, dependency direction, and ownership rules.
- Analyzer authoring: adding and registering analyzers.
- Plugin authoring: external plugin files and built-in extension points.
- Core migration: migration to the engine and API layers.
- Backwards compatibility: stable CLI, config, and output contracts.
- Type-aware analysis: semantic sidecar boundaries, evidence policy, packaging, and corpus validation.
- Similar-code analysis: local semantic discovery, inspect and verdict flow, privacy boundary, packaging, and version 1 scope.
- Implementation references: subsystem-specific durable detail extracted from runtime rules.
- CLI internals: command plumbing, reports, fixes, and format-specific invariants.
- Detection internals: extraction, graph, and analysis behavior.
- Extraction internals: parser dispatch, visitors, component files, and cache-sensitive extraction.
- LSP internals: diagnostics, code actions, and protocol-specific constraints.
- MCP internals: tool contracts, CLI fallbacks, and subprocess behavior.
- Plugin internals: framework plugin discovery and configuration details.
- VS Code internals: extension architecture, binary lifecycle, and editor integration details.
- Drift contract: rules that keep commands and surfaces in agreement, and the harness that checks them.
- Benchmarking: performance measurement and benchmark maintenance.
- Public config corpus: safe public configuration research.
- Security agent verification: verification of syntactic security candidates.
- Fallow compliance: compliance workflow.
- Environment variables: supported environment controls.
- Telemetry: opt-in telemetry behavior and privacy.
- Styling corpus smoke: corpus-wide styling checks.
- Styling PR smoke: pull-request validation.
- Styling release matrix: release coverage across styling surfaces.
Store stable facts, architecture, and operating procedures under docs/.
Store repeatable agent workflows under .agents/skills/. Keep AGENTS.md and
CLAUDE.md as compact routers. Workflow scratch state belongs in the ignored
.plans/ directory and must not become a second documentation tree.