Skill definition

hi-repo-recon

Map an unfamiliar repository into an evidence-backed module inventory, entry-point map, and optional key-function catalog. Use for repository onboarding, architecture reconnaissance, handover preparation, or refactor scoping.

Documentation Architecture

SKILL.md

Repository Reconnaissance

Read shared/leaf-contract.md. Produce structural evidence only. Do not own build auditing, risk synthesis, or full discovery orchestration.

Inputs

Require an absolute, readable repo_root. Accept scope: backend, frontend, infra, data, or all (default). Accept depth: quick, standard (default), or deep. Accept output root; default discovery-output/recon-data.

Reject paths outside the requested repository and output root. Return a structured prerequisite failure when the repository cannot be read.

Retrieval

Read shared/retrieval-protocol.md. Follow the first successful layer and stop when evidence is sufficient. Treat semantic hits as candidates until direct repository evidence confirms them.

Workflow

Validate inputs and create the output directory.
Record retrieval coverage and evidence gaps.
Identify repository roots, buildable/deployable units, source modules, generated/vendor boundaries, and ownership clues.
Identify CLI, API, UI route, service, worker, scheduled-job, and library entry points in scope.
For standard, add key symbols per module. For deep, add verified dependency, IPC, database, external-service, and file-I/O boundaries.
Cross-check indexed results against the filesystem. Report discrepancies; never invent unindexed modules.
Redact secrets, tokens, contact data, connection strings, and infrastructure endpoints.
Write artifacts and manifest per output-contract.md.
Validate required files and JSON shape before returning.

Outputs

Write beneath discovery-output/recon-data/ by default:

module-inventory.json — required.
entry-point-map.json — required.
key-functions.json — required for standard and deep.
artifact-manifest.json — required common envelope.

Complete only when required JSON parses, every claim carries evidence or explicit uncertainty, output paths remain in scope, and the manifest status matches artifacts produced.

If synthesis or build evidence is needed, return the missing prerequisite so the root can select hi-tech-build-audit, hi-module-summary-report, or an orchestrator.

Continue exploring

hi-maintenance-docsOrchestrate a complete, navigable, evidence-backed maintenance documentation pack spanning system overview, architecture, modules, use cases, technical behavior, API contracts, data models, command specifications, gaps, and freshness. Use for end-to-end documentation — not one bounded domain.hi-repository-searchGather traceable evidence from repository code and project documents. Use for codebase exploration, architecture, feature tracing, dependency or impact analysis, and questions that require verified source context.hi-repowikiGenerate or update a comprehensive, evidence-backed repository wiki in Repo Wiki format (/repowiki/), including architecture, installation and configuration, CLI/build automation, languages and frameworks, data/graph schemas, protocol integrations such as MCP, query/search, incremental processing, APIs, testing, deployment, developer guidance, troubleshooting, module documentation, and knowledge cards. Use for codebase documentation, team onboarding, repository capability mapping, and persistent architecture knowledge.