archwaysUnderstand every step.
Install the tools, follow a complete example, and find out exactly what they read, write, and check.
These manuals cover commands, generated files, configuration, failure cases, and implementation details. Examples are identified as illustrative or captured output. Each manual records the source revision it describes.
Generate a codebase reference. Define, plan, and implement changes using that context.
- Capstone user manual
Choose an entry point and find every command in the manual.
- Install and invoke Capstone
Install all skills, invoke commands in your agent, update, and locate global settings.
- Map an existing repository and refresh it
Map existing code, choose refresh/check/rebuild, and understand extraction and staleness.
- Generated files, schemas, and lifecycle
Understand every generated file, required schemas, consumers, Git policy, and retention.
- Take a feature from idea to code
Groom, plan, implement, review, absorb, and resume an existing-product feature.
- Design a new product: start, mockup, and logic
Run start, define a mockup, specify business scenarios, and collect outside answers.
- Choose the design, architecture, standards, and stack
Choose UI/UX, architecture, coding standards, packages, and services; reconcile decisions.
- Build the designed product
Approve and execute the whole-product implementation plan, including brand assets and resume.
- Configure Capstone and monorepos
Configure all supported keys, global/project precedence, headless behavior, and workspaces.
- Interfaces, payload contracts, and Quarry
Write exact edge/payload schemas and feed cross-repository constraints into implementation.
- Freshness checks, schema checks, and CI
Run literal check commands, interpret output and exits, install CI, and understand limits.
- Review code, repair documentation, and improve instructions
Review code and UI, repair documentation state, and approve evidence-backed instruction edits.
- Version notes and less obvious behaviors
Resolve upgrades and obscure behavior, including source discrepancies, hashes, index paths, and lifecycle traps.
Publish documentation across repositories. Search sections, trace dependencies, and compare contract fields.
- Quarry documentation
A source-verified practical manual for installing Quarry, importing docs, querying the graph, checking contracts, and operating CI.
- Quarry: what it does and how to install it
Understand the source repo, shared docs repo, local clone, and embedded index; install or build the binary.
- First import and recorded walkthrough
Create the minimum documentation, register a repository, inspect the shared estate, and distinguish recorded output from examples.
- Configuration and monorepo targets
Every config key and environment variable, path validation, target registration, umbrella imports, and target retirement.
- Imports, generated files, ownership, and concurrency
What add/update/remove actually write, stamp ordering, strict checks, permalink rewriting, and concurrent-push limits.
- Writing pages, interfaces, aliases, and observed edges
The complete Markdown/frontmatter contract, three declaration forms, name joins, alias precedence, and traffic input schema.
- Reading, searching, and traversing dependencies
Every docs query with examples, exact matching rules, graph directions, shortest paths, and confidence markers.
- The SQLite index, SQL, and freshness
Exact schema, rebuild triggers and transaction, FTS query SQL, read-only inspection recipes, cache caveats, and four freshness clocks.
- Checking documented contracts before a merge
Producer working-tree checks, consumer snapshots, inline and model-backed field formats, verdicts, missing-data behavior, and limitations.
- CI, publishing workflows, and repository audits
Wire default-branch imports, a pull-request check, the reusable GitHub App workflow, scheduled audits, and automation diagnostics.
- Command, output, and exit-code reference
All verbs and flags, machine-readable envelope schemas, payload fields, recorded examples, and parser-stage exceptions.
- Troubleshooting and recovery
Diagnose setup, identity, imports, empty answers, contract coverage, cache state, credentials, and busy-repository failures.
- Parser details and implementation limits
Less obvious implementation facts: JSON/YAML coercion, identity loss, path handling, table heuristics, freshness gaps, retries, and detector patterns.