Skip to main content
Requirements traceability

IntentSutra AI

Local requirements-to-evidence. Not a completion certificate.

Local requirements-to-evidence traceability. npm [email protected] published 2026-11-22. The command aliases are intentsutra and intentsutra-ai. The npm README says there is no API key, no cloud, no telemetry, and 0 runtime dependencies. It gives no legal advice and certifies nothing. It does not run tests. An imported test or release result is labelled not independently verified. A local reviewer is self-attested. It never marks a requirement completed. A ticket, pull request, test file, or imported status is a pointer, not proof. On 2026-11-23 the npm downloads API returned no row, so both counts are 0. GitHub showed 0 stars and 0 open issues, and its detected language was HTML because the handbook is in the repository. The CLI file is packages/cli/index.mjs. The handbook home said 116 passed, 2 skipped, 0 failed, recorded by npm run proof on 2026-11-22T18:35:02.198Z on Node v24.15.0. The npm README does not state a test count. This page did not re-run the tests. The name IntentSutra has not been trademark-searched. The npm README says the project is not affiliated with an IDE vendor, a model provider, a standards body, or a regulator.

  • Offline
  • 11 checks
  • Node 22+
GitHub version
0.2.1
Licence
Apache-2.0
Language
JavaScript
Weekly downloads
0
Overall downloads
0
GitHub stars
0

Use it when

  • 1

    You want stable requirement IDs and explicit unknowns from a local JSON or Markdown file.

  • 2

    You want a named person to approve one exact revision, with the approval going stale when that revision changes.

  • 3

    You want an independent audit report in JSON, Markdown, and HTML without an API key.

What it does, and what it does not

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

Does

  • ingest stores stable requirement IDs and a SHA-256 of each source in .intentsutra/requirements.json.
  • review lists missing actor, trigger, constraints, authorization, delivery, and money as questions. The npm README says it does not invent those answers. It also writes five candidate Gherkin scenarios marked unapproved.
  • approve records a named reviewer, a note, and the exact revision, and it requires --confirm-human. The npm README says that identity is self-attested.
  • link binds code, test, test-run, release, ticket, or pull-request evidence to one revision and hashes the files. revise and impact keep history and mark old approvals and links stale.
  • attest can return human-attested-evidence-linked. The npm README says that word is never completed.
  • audit runs 11 checks and writes a sealed JSON, Markdown, and HTML report. verify checks the seal. A hash-chained log is written to .intentsutra/audit.jsonl. serve binds 127.0.0.1, requires a bearer token of 24 or more characters, and accepts GET only. Node.js 22 or newer. 0 runtime dependencies.

Does not

  • It does not give legal advice and it certifies nothing. The handbook legal page says it is informational and tells the reader to ask a qualified lawyer.
  • It does not run the project tests. The npm README says imported test and release outcomes are labelled not independently verified.
  • It never marks a requirement completed. The npm README says a ticket, a pull request, a test file, or an imported status is a pointer, not proof.
  • A hash and the audit chain detect edits. The npm README says they are not signatures, and a person with write access can rewrite the files.
  • install-ide copies instruction files for nine IDE hosts and refuses to overwrite without --force. The npm README says native activation in each host is not verified. The files are instructions, not running services.
  • The npm README lists real test-execution capture, verified identity, a signed audit, a dashboard, a database, an MCP server, and connectors as planned and not built. The handbook home said 116 passed, 2 skipped, and 0 failed. This page did not re-run them and did not re-run the sample audit. The name IntentSutra has not been trademark-searched.

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 IntentSutra AI npm README, version 0.2.0, published 2026-11-22. It reads files you pass in. It does not decide that a requirement is done.

  1. 1. Ingest

    Stable IDs

    Each source is hashed. Missing facts become questions. The npm README says those answers are not invented.

  2. 2. Approve

    One revision

    A named reviewer, a note, and --confirm-human. The npm README says the identity is self-attested.

  3. 3. Link

    Hashed evidence

    Code, tests, tickets, and releases point at that revision. A change marks the old approval stale.

  4. 4. Audit

    11 checks

    Sealed JSON, Markdown, and HTML. The npm README says a silent edit of the sample workspace returns FAIL. The seal is not a signature.

How to read the result

Result It means It does not mean
Unknown The npm README says a missing fact stays a question and a [TO CONFIRM] placeholder. Not an invented answer and not an approval.
human-attested-evidence-linked The npm README says this is the strongest attest result. Not completed. A pointer is not proof.
FAIL The npm README says a silent edit of the sample audit workspace returns this verdict. Not a legal finding. This page did not re-run that sample.

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] doctor local check. No network.
  • intentsutra init start a workspace.
  • intentsutra ingest --file team-invitations.json stable IDs from a local file.
  • intentsutra audit --out audit-out JSON, Markdown, and HTML. verify checks the seal.
  • review, approve, link, revise, impact, attest, trace, report, install-ide, and serve listed in the npm README. serve is loopback only.

Questions

Does IntentSutra prove a requirement is complete?

No. The npm README says it never marks a requirement completed. A ticket, a pull request, a test file, or an imported status is a pointer, not proof. Imported test results are not independently verified. This page is not legal advice either.

Where is the handbook, and did this page re-run the tests?

https://shahvatsalofficial.github.io/intentsutra-ai/ returned HTTP 200 on 2026-11-23. Fourteen handbook pages share one meta description, so this page shows that sentence on the documentation home card only. samples/audit/audit-report.html and samples/audit-forged/audit-report.html also returned HTTP 200, share the HTML title IntentSutra AI — Audit Report, and have no meta description. test-results/test-run.tap returned HTTP 200 and is not HTML, so that card uses the file name. The handbook home said 116 passed, 2 skipped, and 0 failed, recorded by npm run proof on 2026-11-22 on Node v24.15.0. The npm README does not state a test count. This page did not re-run the tests and did not re-run either sample audit.

WhatsApp