Tracewell MCP Server
Publishes release trace reports to Confluence, including the readiness verdict, coverage, blockers, trace matrix, and risk-control table with Jira issue links. Republishing for the same release adds a page version, using Confluence page history as an audit trail.
Provides design-control traceability for Jira issues: builds trace chains from issue types, labels, and links; detects missing downstream trace, orphans, failed or unrun tests, and suspect test results; analyzes change impact; checks release readiness; comments on affected issues for suspect-link automation; and stores per-project trace settings as Jira project properties.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Tracewell
Design-control traceability for Jira and Confluence, built on Atlassian Forge with a Rovo agent and an MCP server.
Teams that build regulated products (medical device software, automotive, aerospace) must prove that every user need is met by requirements, every requirement is implemented and verified, and every hazard is controlled and the control verified. Most teams keep that proof in spreadsheets that go stale the moment a requirement changes. Tracewell reads the trace straight from Jira, finds the gaps, flags tests that passed before their requirement changed, and publishes the trace report to Confluence.
What it does
Trace model. User needs, design inputs, design outputs, verification, validation, hazards and risk controls, mapped from Jira issue types or
tw-labels. Trace links are ordinary Jira issue links; direction comes from the design-control chain, so it does not matter which way a link was drawn.Gap rules. Missing downstream trace, orphans, failed tests, tests not run, and suspect results (a test that passed before anything upstream of it changed). Only content edits count; status moves and comments do not.
Suspect-link automation. A Jira trigger watches traced items. When the summary, description, acceptance criteria or labels of a requirement change, Tracewell comments on every passed test below it ("suspect result, re-run before release") and on the requirement itself with the full re-test list.
Change impact. For any item: tests to re-run, items to review, hazards involved and the user needs at stake.
Release readiness. For a fix version: the items in the release plus their full chains, coverage, blockers and warnings, with a READY / NOT READY verdict.
Risk controls. Hazard to risk control to verification table, with each hazard marked controlled, partial or open.
Confluence trace report. One click (or one Rovo request) publishes the release report with the verdict, coverage, blockers, trace matrix and risk table, with issue keys linked back to Jira. Publishing again for the same release adds a page version, so Confluence page history becomes the audit trail.
Per-project settings. A Jira project settings page maps your own issue types, labels and link types, and sets which trace links are required. Settings are validated and stored as a Jira project property.
Related MCP server: JIRA MCP Server
Forge modules
Module | Use |
| Trace chain, gaps, suspect warning and change impact on every traced issue, plus an "Ask Tracewell" button that opens the Rovo agent |
| Tabs for overview (coverage, gap chart), trace matrix with CSV, risk table and release check with Confluence publish |
| Project configuration |
| Suspect-link automation |
| The Tracewell agent with conversation starters |
|
|
|
|
| Exposes the five read-only actions as tools on the Atlassian MCP server |
Scopes: read:jira-work, write:jira-work (comments and the settings property), read:space:confluence, read:page:confluence, write:page:confluence. No egress: Tracewell calls only Atlassian APIs.
Links
Install (judges)
Open the installation link and install Tracewell on your Jira and Confluence site (it asks for both).
In any Jira project, create a few issues with Tracewell labels, or use issue types with these names:
Kind
Issue type names
Label
User need
User Need
tw-user-needDesign input
Requirement, Design Input
tw-design-inputDesign output
Design Output, Specification
tw-design-outputVerification
Test Case, Verification
tw-verificationValidation
Validation
tw-validationHazard
Hazard, Risk
tw-hazardRisk control
Risk Control, Mitigation
tw-risk-controlLink them with the built-in Relates link (or Traces, Verifies, Mitigates, Implements). Mark a test as failed with the
tw-failedlabel; a test passes when it is moved to a done status. Set fix versions to group items into a release.Or seed the full demo project in one command (needs a Jira API token for your own site; the token is read from the environment and never printed):
JIRA_SITE=https://your-site.atlassian.net JIRA_EMAIL=you@example.com JIRA_API_TOKEN=... node scripts/seed-demo.ts MEDOpen the project sidebar item Tracewell, or any traced issue, or ask the Tracewell agent in Rovo: "Is version 2.1.0 ready to release?"
To see suspect-link automation, edit the description of a requirement that has a passed test below it (or run
node scripts/seed-demo.ts MED --edit <key>), then open the test.
Develop
pnpm install
pnpm test # 92 vitest tests, fully offline against an in-memory Jira and Confluence
# plus an opt-in live smoke test: see test/live.test.ts
pnpm typecheck
pnpm validate:manifest # the Forge CLI's own manifest validator, offline
forge deploy && forge installAll trace logic is in src/lib as pure TypeScript with no Forge imports. src/index.ts is the only file that touches the Forge runtime.
Limits
Tracewell checks traceability. It does not certify compliance with ISO 13485, IEC 62304, ISO 14971 or FDA 21 CFR 820, and it does not replace a quality reviewer. One project load reads up to 1,000 traced issues.
Built with AI coding assistance; every rule is covered by tests.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Read-only AI coding tools for change verification, release readiness, capacity, and guidance.
Read-only MCP access to a documented IT fleet: state, changes, posture. 15 tools.
A read-only verified record of agent-operable GTM tools: search, fetch, compare, track changes.
Read-only XRP Ledger token-risk, issuer-authenticity and wallet-security checks.
Related MCP Servers
- AlicenseBqualityDmaintenanceProvides tools for AI assistants to interact with JIRA APIs, enabling them to read, create, update, and manage JIRA issues through standardized MCP tools.610 npm3MIT
- AlicenseAqualityDmaintenanceProvides read-only access to JIRA REST API, enabling LLMs to query and retrieve information from JIRA instances.1414 npmMIT
- AlicenseAqualityDmaintenanceExposes generic Jira REST tools and native Jira Product Discovery operations for Jira Cloud, including insights and attachments.8156 npmISC
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to deeply introspect Jira Data Center configurations — including workflows, schemes, automation, and Assets — through 76 read-only tools, without any modification capability.6MIT