Skip to content

CLAUDE.md

Latest commit

 

History

History
122 lines (100 loc) · 5.64 KB

CLAUDE.md

File metadata and controls

122 lines (100 loc) · 5.64 KB
Raw

OpenCodeHub MCP Tools

This repository has been indexed by OpenCodeHub. When you are working in this codebase, prefer the following MCP tools over raw file search — they return graph-aware results grouped by execution flow and include blast-radius risk tiers.

  • list_repos — enumerate repos currently indexed on this machine.
  • query — hybrid BM25 + vector search over symbols, grouped by process.
  • context — inbound/outbound refs and participating flows for one symbol.
  • impact — dependents of a target up to a configurable depth, with a risk tier.
  • detect_changes — map an uncommitted or committed diff to affected symbols.
  • rename — graph-assisted multi-file rename; dry-run is the default.
  • sql — read-only SQL against the local graph store with a 5 s timeout.

Run codehub analyze after pulling new commits so the index stays aligned with the working tree. codehub status reports staleness.

Full MCP surface

The full MCP surface is 28 tools (see packages/mcp/src/server.ts); the 7 listed above are the high-frequency exploration tools. For the full inventory, use the /opencodehub-guide skill.

AMBIGUOUS_REPO

When two or more repos are indexed on this machine, per-repo tools require an explicit repo: (or the repo_uri: alias — a Sourcegraph-style URI such as github.com/org/repo, or local:<hash> for unpublished repos) and return AMBIGUOUS_REPO otherwise. The error envelope carries a structured _meta payload on structuredContent.error: { error_code: "AMBIGUOUS_REPO", jsonrpc_code: -32602, choices: [ { repo_uri, default_branch, group } ] (capped at 10), total_matches, hint } — so the calling agent can retry deterministically with a single repo_uri from choices. When total_matches > choices.length, the caller knows the list was truncated.

See ADR 0012 (docs/adr/0012-repo-as-first-class-node.md) for the rationale behind repo_uri as a first-class node attribute. The repo_uri shape was promoted to a typed graph attribute by AC-M6-1 (packages/core-types/src/nodes.ts:524-552). group_cross_repo_links (the AC-M6-3-reframed MCP tool) and the group_* family (AC-M6-4) all emit repo_uri in the same canonical form, so a caller can use any of those tools' repo_uri outputs as input to AMBIGUOUS_REPO.choices retries.

Worked example — error envelope, then retry:

// Error envelope returned by a per-repo tool when two repos are indexed
{
  "structuredContent": {
    "error": {
      "error_code": "AMBIGUOUS_REPO",
      "jsonrpc_code": -32602,
      "choices": [
        { "repo_uri": "github.com/org/api-svc", "default_branch": "main", "group": "platform" },
        { "repo_uri": "github.com/org/billing-svc", "default_branch": "main", "group": "platform" }
      ],
      "total_matches": 2,
      "hint": "Retry with repo_uri=<one of above>"
    }
  }
}
// Retry — pick the first choice deterministically
{ "tool": "context", "args": { "repo_uri": "github.com/org/api-svc", "symbol": "..." } }

Durable lessons

Prior-session architecture lessons live in .erpaval/INDEX.md (SCIP edge conventions, BM25 caveats, SageMaker embedder patterns). Read before making graph-index or retrieval changes.

Claude Code plugin

This repo ships a Claude Code plugin at plugins/opencodehub/ — it provides a code-analyst subagent and 10 skills. Install via codehub init (writes .mcp.json + links the plugin).

Storage backend — lbug graph + DuckDB temporal

The graph tier is always @ladybugdb/core (graph.lbug); the temporal tier — cochanges, structured symbol summaries, and the codehub query --sql escape hatch — is always DuckDB (temporal.duckdb). Both files live under <repo>/.codehub/. There is no env-var, no probe, no fallback; if the lbug binding fails to load, open() throws GraphDbBindingError and the operation aborts. See ADR 0016 (docs/adr/0016-duckdb-graph-rip.md) for the rationale and the AGE/Memgraph/Neo4j/Neptune community-adapter contract that survives the rip-out (the segregated IGraphStore / ITemporalStore interfaces stay exactly because community-fork adapters are a deliberate escape hatch).

IGraphStore lives only on GraphDbStore; DuckDbStore implements ITemporalStore only. Embeddings live in graph.lbug and stream into a per-call DuckDB temp table at pack time so the byte-identical Parquet sidecar still works (see packages/pack/src/embeddings-sidecar.ts). Future temporal swap (e.g. SQLite-WASM) only needs a new ITemporalStore implementor — no graph-tier change.

Parse runtime — WASM-only, vendored grammars

@opencodehub/ingestion runs web-tree-sitter (WASM) as the only parse runtime on Node 20, 22, and 24. There is no native opt-in — the legacy parser-runtime env var and CLI flag were removed in 0.4.0 (see ADR 0015 and the root + per-package CHANGELOGs). The CLI continues to emit a one-shot stderr advisory if a stale env var is set, then ignores it.

All 15 GA grammar .wasm blobs are vendored at packages/ingestion/vendor/wasms/, built from the grammar sources pinned in package.json. Rebuild via bash scripts/build-vendor-wasms.sh after bumping any of those grammars — requires docker, podman, finch (aliased as docker), or a local emcc install. Re-vendoring is a one-shot operation; consumers never build grammars at install time.

The complexity phase (packages/ingestion/src/pipeline/phases/complexity.ts) has been ported to web-tree-sitter, so cyclomatic-complexity metrics run on every install with no native dependency at runtime or test time. ADR 0013 (docs/adr/0013-parse-runtime-wasm-default.md) is superseded by ADR 0015 (docs/adr/0015-wasm-only-parser-at-the-npm-distributed-boundary.md).