Skip to main content
Open source

ContribSutra AI

Prepare a first contribution. Do not open the pull request.

Offline contributor-onboarding CLI. npm [email protected] published 2026-11-15. The README says it never opens a pull request, never comments, and never runs a repository script. npm last-week downloads were 245 (2026-11-09 to 2026-11-15). The 2026-10-01 to 2026-11-17 total was the same 245. GitHub showed 4 stars and 0 open issues on 2026-11-17.

  • Offline
  • 25 skills
  • Node 20+
GitHub version
0.2.1
Licence
Apache-2.0
Language
JavaScript
Weekly downloads
274
Overall downloads
274
GitHub stars
4

Live npm weekly downloads, npm overall downloads, and the latest npm version, plus GitHub stars and the package.json version, fetched 2026-11-17 19:07 UTC. Cached 10 minutes. Weekly downloads cover 2026-11-10 to 2026-11-16. Overall downloads cover 2026-10-01 to 2026-11-17. npm does not publish an earlier total. Stars and downloads show public use. They are not a quality score.

Use it when

  • 1

    You cloned a repository and want its contributor files mapped with file:line sources.

  • 2

    You exported issues yourself with gh and want closed, assigned, locked, and security-sensitive issues declined.

  • 3

    You want a PR packet that keeps verified, user-reported, inferred, and not-run apart.

What it does, and what it does not

Read this before the guide. A limit here is a limit of the tool.

Does

  • scan maps README, CONTRIBUTING, LICENSE, conduct, security, templates, CODEOWNERS, Node/Python/Rust/Go manifests, CI workflows, and policy signals, with file:line sources and SHA-256.
  • doctor runs fixed, bounded --version probes and compares them with versions the repository declares.
  • issues ranks an offline issue snapshot. It declines closed, assigned, locked, security-sensitive, prompt-injection, and large-scope issues.
  • plan writes six learning steps in guided, collaborative, or self-directed mode. The README labels that PLANNED_NOT_EXECUTED.
  • pr-check writes a packet that keeps verified, user-reported, inferred, and not-run separate. Test claims stay unverified_external_claim.
  • audit runs 30 adversarial safety checks and a negative control. Exit 0 is ok. Exit 1 is audit gaps or failed verification. Exit 2 is a usage or input error.
  • install-skills copies 25 skills and refuses to overwrite. The README says 20 are specialist roles and 5 are capability skills. Copying is not proof the IDE activates them.

Does not

  • It does not open, comment, assign, label, or merge anything on GitHub. Live GitHub read-only API is planned, not implemented.
  • It does not run npm install, make, pip, or any script found in a repository, README, or issue.
  • It does not treat a badge, a person, or a file as proof a test passed.
  • The audit is a self-assessment by the same tool. The README says it is not third-party certification.
  • SHA-256 detects a changed file. The README says it is not a signature. Redaction is heuristic and cannot prove nothing sensitive remains.
  • Sandboxed test execution, Dev Container support, patch analysis, a hosted dashboard, OIDC/RBAC, and draft-PR writes are listed as not implemented. Adoption and acceptance rates are not measured.

Claims above are from the project pages and are attributed as stated there. They are not independently verified here.

How a run moves

From the ContribSutra AI README, npm 0.2.0, published 2026-11-15. It reads a clone and an issue file you exported. It does not open a pull request.

  1. 1. Scan

    Contributor files

    README, CONTRIBUTING, license, conduct, security, templates, and manifests, each with a file:line source.

  2. 2. Rank

    Offline issues

    Closed, assigned, locked, security-sensitive, instruction-like, and very large issues are declined. A label alone is not a welcome.

  3. 3. Plan

    Not executed

    Six learning steps. The README marks the plan PLANNED_NOT_EXECUTED. It does not run the repository tests.

  4. 4. Packet

    You decide

    pr-check separates verified, user-reported, inferred, and not-run. You rewrite the draft and you submit it.

How to read the result

Result It means It does not mean
Declined The issue was closed, assigned, locked, security-sensitive, instruction-like, or too large. Not a claim the maintainer wants a contribution.
PLANNED_NOT_EXECUTED A learning plan was written. Not a patch and not a test run.
Self-assessment audit ran 30 checks against this tool. Not a third-party certificate. The README says you can re-run npm test and compare SHA-256.

Wording is from the project page or README and is attributed as stated there. It is not independently verified here.

Commands and notes

  • npx [email protected] audit --out ./contribsutra-audit 30 adversarial checks on this machine.
  • contribsutra-ai scan --root ./some-repo --out ./reports/scan policies, license, toolchain, file hashes.
  • contribsutra-ai doctor --root ./some-repo declared tool versions versus what is installed.
  • contribsutra-ai issues --issues ./issues.json you export the snapshot with gh. The CLI does not call GitHub.
  • contribsutra-ai plan, pr-check, verify, render, install-skills, and serve listed in the README command table. serve binds 127.0.0.1 only.

Guide

How to use the result. GitHub is the source repository. npm is the install.

What 0.2 checks

From the ContribSutra AI README, npm 0.2.0, published 2026-11-15. Node.js 20 or newer. 0 runtime dependencies.

  • npx [email protected] audit --out ./contribsutra-audit writes audit-report.html, audit-report.md, audit-report.json, and SHA256SUMS.txt. The audit runs 30 adversarial checks.
  • scan reads contributor files on a clone you already have. doctor compares declared tool versions with bounded --version probes.
  • issues ranks a JSON snapshot you export with gh. The CLI does not call GitHub. Closed, assigned, locked, security-sensitive, instruction-like, and very large issues are declined.
  • plan writes six steps and marks them PLANNED_NOT_EXECUTED. pr-check keeps verified, user-reported, inferred, and not-run apart. Exit 0 is ok. Exit 1 is audit gaps or failed verification. Exit 2 is a usage or input error.

What it will not do

  • The README says it never opens, comments, assigns, labels, or merges a pull request, and it never runs npm install, make, pip, or a script from the repository.
  • The audit is a self-assessment, not a third-party certificate. SHA-256 detects a changed file and is not a signature.
  • Live GitHub API, sandboxed tests, Dev Containers, patch analysis, a hosted dashboard, and draft-PR writes are not implemented. Adoption is not measured.
  • 25 skills are copied and existing files are not overwritten. IDE activation is not verified. No telemetry.

Questions

Will ContribSutra open my pull request?

No. The README says it never opens, comments, assigns, labels, or merges anything on GitHub, and it will not become a PR-generating bot. You write the change and you decide whether to submit.

Does the audit prove the tool is certified?

No. The README says the audit is a self-assessment with reproducible evidence, not third-party certification. SHA-256 detects a changed file and is not a signature. The README points at docs/proof for the recorded test and fuzz counts. This page does not repeat those counts as if they were re-run here.

WhatsApp