Career Evidence
Get started
LOCAL-FIRST. AI OPTIONAL.

Your work tells a story.
Make it visible.

Bring tasks, commits, and pull requests together into a structured record of what you've built — locally, with or without AI.

Get started Read the documentation
npm install -g @career-evidence/cli

Node.js 20+ · Git · No AI key required

127.0.0.1:4310YOUR LOCAL WORKSPACE
Career Evidence local Web UI showing an example project, task summaries, and evidence correlation review.Career Evidence local Web UI showing an example project, task summaries, and evidence correlation review.
ce run platform Evidence collected → JSON exportedRuns locally
From task to implementation. With the context intact.Actual Web UI · Synthetic example data
BUILT FOR THE WAY YOU BUILD
TasksGit historyEvidenceYour story

01 / THE MISSING CONTEXT

Your work is everywhere.
Your story shouldn’t be.

The task is in a tracker.

Requirements, decisions, and discussions spread across ClickUp, Linear, or Jira.

The implementation is in Git.

The changes span branches, commits, pull requests, and more than one repository.

The details get harder to find.

Preparing a work summary becomes an exercise in reconstructing your own history.

Career Evidence brings the pieces together.

02 / A CONNECTED WORKFLOW

From scattered activity
to structured evidence.

  1. 01

    Connect your sources

    Choose your task tracker and the repositories that hold the implementation.

    Task sourceLinear
    Repositoriesapi / web
  2. 02

    Collect and analyze

    Reconstruct task-related history from available evidence. Add AI analysis if you choose.

    $ ce run my-project Tasks collected Git evidence reconstructed
  3. 03

    Explore and export

    Review associations, add your context, and keep structured results you can return to.

    all-tasks.jsonJSON
    task-summaries.jsonJSON

03 / THE WORK, IN CONTEXT

See the work
behind every task.

Start with a task. Follow its implementation.
Keep the evidence and your interpretation distinct.

Actual Career Evidence task editor with summary, personal role, key changes, and manual context fields, populated with synthetic data.Actual Career Evidence task editor with summary, personal role, key changes, and manual context fields, populated with synthetic data.
01 / EXPLORE

More than a
list of tickets.

Search task summaries, revisit technical changes, and add the details that only you can provide.

  • Task summaries and technical notes
  • Explicit unknowns and analysis status
  • Durable manual context
Actual Web UI · Synthetic data
Actual Career Evidence correlation review showing synthetic branch and pull request candidates with Confirm and Exclude actions.Actual Career Evidence correlation review showing synthetic branch and pull request candidates with Confirm and Exclude actions.
02 / REVIEW

A link is a starting
point for review.

Inspect suggested branch, commit, and PR associations. Confirm useful matches or exclude misleading ones.

  • Source identities and matching reasons
  • Manual Confirm, Exclude, and Reset
  • Reviewable decisions, not ownership claims
Actual Web UI · Synthetic data
task-summaries.jsonJSON
{
  "taskId": "ENG-142",
  "title": "Add retry handling to the API client",
  "status": "success",
  "summary": "Added bounded retries and connection recovery.",
  "signals": [],
  "role": "Unable to determine from available evidence.",
  "results": [],
  "areas": [
    "backend"
  ],
  "keyChanges": [
    "Add bounded retries for transient errors",
    "Handle connection recovery"
  ],
  "unknowns": [
    "Personal ownership requires manual context."
  ],
  "details": {}
}
Illustrative excerpt · Synthetic data
03 / EXPORT

Useful beyond
the dashboard.

Keep detailed source evidence and compact summaries in JSON. Revisit your history or use it in your own workflow.

  • all-tasks.json for detailed evidence
  • task-summaries.json for summaries
  • Manual details preserved on refresh
Explore the output format

Evidence depends on available history and source access. Suggested associations and analysis need your review.

04 / MADE FOR REAL ENGINEERING

Built around how developers
actually work.

Multiple task sources

Bring ClickUp, Linear, or Jira Cloud tasks into one consistent evidence workflow.

Multi-repository context

Follow implementation across related repositories, with existing checkouts or managed clones.

Pull request context

Enrich local Git history with available GitHub PRs, GitLab MRs, or Bitbucket Cloud PRs.

Optional AI analysis

Choose a supported model for technical summaries, or use conservative, deterministic analysis.

Local-first workflow

Your CLI, dashboard, configuration, and evidence files run in your own environment.

Structured exports

Keep detailed evidence and compact task summaries in portable JSON files.

05 / CONNECT THE CONTEXT

Works with the tools
you already use.

Choose your sources.
Keep your existing workflow.

Analysis, your way

REMOTE PROVIDERS

OpenAIGoogle GeminixAIMistralDeepSeekGroqOpenRouterAlibaba Model Studio

LOCAL / CUSTOM ENDPOINTS

Ollama · OpenAI-compatible
No-AI mode

Deterministic analysis. No model required.

All integrations shown are implemented. Access depends on provider permissions, configuration, and available history.

06 / LOCAL BY DESIGN

Your workflow.
Your environment.
Your choice.

Run the CLI and dashboard on your machine. Keep your evidence in local files. Choose whether a model belongs in the process.

  • CLI and Web UI run locally
  • Remote AI, local AI, or no AI
  • JSON files you can inspect and keep
Understand the data flow
your-machine — zsh
# A no-AI workflow, after configuration$ ce run my-project
Collect tasks from configured sourceReconstruct local Git evidenceGenerate deterministic summariesWrite JSON exports
$ ce webhttp://127.0.0.1:4310Illustrative output · Real commands
AI provider: none

Local-first describes where the tool runs and stores results. Task sources, enabled SCM APIs, and remote AI require network access and receive the data needed for their functions. Diff modes may send selected code hunks.

07 / START WITH YOUR SOURCES

Get started
from your terminal.

One package. Your task tracker.
Your existing Git checkout.

Read the complete guide
ClickUp + local Git
No AI or SCM API token needed
01

Install and initialize

Requires Node.js 20+, npm, and Git.

npm install -g @career-evidence/cli
career-evidence init
career-evidence config path
02

Configure your source and repository

Replace the generated config with the no-AI example below. Add your ClickUp token, workspace and user IDs, absolute checkout path, and base branch.

The init template enables remote AI and GitHub enrichment. Replace it before running this no-AI workflow.

Show complete no-AI configuration
{
  "projects": {
    "my-project": {
      "taskSource": {
        "provider": "clickup",
        "apiKey": "<clickup-personal-api-token>",
        "workspaceId": "123456",
        "userIds": [
          "789"
        ]
      },
      "scm": {
        "provider": "github",
        "enabled": false,
        "apiKey": "",
        "repositories": [
          {
            "name": "application",
            "localPath": "/absolute/path/to/your/repository",
            "baseBranches": [
              "main"
            ],
            "enrichment": false
          }
        ]
      },
      "analysis": {
        "enabled": false
      }
    }
  }
}

Values in angle brackets are placeholders. Protect your config: credentials are stored as plaintext. Prepare access

03

Collect, analyze, and open

Validate configuration, run the pipeline, then open the local URL printed by web.

career-evidence config validate my-project
career-evidence run my-project
career-evidence web

08 / A FEW USEFUL ANSWERS

Good questions.
Clear answers.

The details, before you get started.

What is Career Evidence?

A local-first developer tool that connects task trackers with Git history to reconstruct a structured, traceable record of engineering work. Use the CLI and local Web UI to explore tasks, review suggested links, add context, and export results. It does not automatically establish personal ownership or business impact.

Does it require an AI API key?

No. Deterministic analysis needs no AI key. Local AI can use Ollama or an OpenAI-compatible local endpoint. Remote AI requires the credentials and model access expected by your chosen provider. Task trackers still require their own credentials.

Can I use it without AI?

Yes. Set analysis to exactly { "enabled": false } and keep the default buildIn evidence ranking. Collection, conservative summaries, manual review, editing, and JSON exports remain available. The generated init template enables remote AI, so replace it with the documented no-AI configuration before running.

Does it work with multiple repositories?

Yes. Configure several repositories per project, using existing local checkouts or managed clones. Career Evidence collects available Git evidence across them. One task source and one SCM API provider can be selected per project.

Which task management systems are supported?

ClickUp, Linear, and Jira Cloud. Collection is limited by account permissions, source visibility, and your configured scope and assignee filters. Jira Server and Data Center are not supported.

Does the Web UI run locally?

Yes. Run career-evidence web and open the URL printed in your terminal, normally http://127.0.0.1:4310. The dashboard is bundled with the CLI and binds to the loopback interface. No separate frontend server or database is needed.

Where is my data processed?

Configuration, evidence, summaries, and caches are stored locally. Task collection and enabled SCM enrichment contact the configured services. Remote AI receives the context needed for analysis; diff modes may send selected code hunks. A local model uses your configured endpoint. Local storage is not encrypted by the application.

Can I export the results?

Yes. run or export writes all-tasks.json with detailed evidence and analysis, and task-summaries.json with compact summaries and manual details. Exports are available without AI. Review matching decisions and unknowns before sharing; successful analysis does not guarantee complete reconstruction.

Make your work
easier to understand.

Connect your engineering history and start building
a clearer record of what you’ve delivered.

Get startedExplore the npm package
Local-first. AI optional.