The Repository Metadata Parser
Parse a local repository’s dependency files — package.json, requirements.txt, go.mod, Cargo.toml and more — to infer its languages, frameworks, and build commands, saved to file for downstream agents, on infrastructure you control.
The facts about a project are scattered across files
A repository’s essential metadata — its languages, frameworks, and how to build it — is spread across half a dozen manifest formats. Collecting it by hand is tedious and inconsistent across ecosystems.
Many formats
Each ecosystem has its own manifest, with its own conventions.
Build steps are implicit
How to actually build the project isn’t obvious from the files.
No reusable artifact
What’s learned by hand isn’t saved for the next step or tool.
Inconsistent results
Manual reads vary by who does them.
Manifests in, metadata out
Parsing
Every common manifest
One parser, many formats.
The tool parses dependency files across ecosystems — package.json, requirements.txt, go.mod, Cargo.toml and more — to inventory what a project declares, consistently regardless of language.
- Multi-format manifest parsing
- Language and framework inference
- Cross-ecosystem
- Local-repo input
One parser
Inference
Build commands and stack
How to actually run it.
From the manifests it infers the languages, frameworks, and the build commands, then writes the result to file so downstream agents and tools can consume a clean, reusable artifact.
Saved to file
Governance
On-premise parsing
Your files, your control.
Parsing runs against a local repository inside your perimeter with audit logging, writing output to a directory you choose.
Local, logged
Parameters
The repo_metadata tool accepts these inputs when an agent calls it. Required inputs are flagged.
default: /tmp/repo_metadata/ Optional Directory where the metadata file is saved.
Where metadata parsing pays back
Pipeline grounding
Produce a metadata artifact other tools build on.
Build discovery
Infer how to build a project automatically.
Portfolio inventory
Catalog the stack of every repo from its manifests.
CI setup
Seed CI configuration from inferred build commands.
Documentation
Feed accurate metadata into generated docs.
Agent workflows
Give downstream agents a clean, reusable project profile.
Assigned to agents, orchestrated as networks
On VDF AI, an industry’s use cases map to agents, and you assign tools like this one to those agents. Compose multiple agents into a governed, on-premise network.
What changes after you assign it
Questions about the Repository Metadata Parser tool
What does the repository metadata parser do?
It parses a local repository’s dependency files across ecosystems — package.json, requirements.txt, go.mod, Cargo.toml and more — to infer languages, frameworks, and build commands, saving the result to a file for downstream use.
How is it different from the tech stack detector?
The detector works on a GitHub snapshot and reports the stack; this tool works on a local repository, parses the manifests directly, infers build commands, and writes a reusable artifact to file.
Where does it save output?
To the output_dir you specify, defaulting to a temporary directory, so other tools and agents can consume the metadata.
Is our code exposed?
No. Parsing runs on-premise against a local repository with audit logging.
How does it pair with other tools?
It produces input for the dependency graph builder, the doc generators, and CI-setup workflows that need a clean project profile.
Tools that work well alongside this one
Where this tool delivers value
Turn manifests into a clean project profile
See the repository metadata parser feed a build pipeline for an agent — on infrastructure you control.