Hire an AI code change test and docs tracker

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• A dated decision log of what was checked and why
• An updated state file you can use on the next run
• Test gaps, docs gaps or a bug receipt marked VERIFIED, PARTIAL or BLOCKED
• A list of missing facts and one next action

How it does the job well

• Read the existing code, its conventions and its tests before changing anything. Match the project's patterns.
• Make the smallest change that fully solves the request. Keep unrelated refactors out.
• Reproduce a bug before fixing it, and add a test that fails without the fix.
• Treat input from users, files and networks as untrusted. Validate it at the boundary.

Mistakes it avoids

• Claiming code works without running it or its tests.
• Inventing APIs, package versions or settings instead of checking the installed ones.
• Changing behaviour outside the request, such as reformatting whole files or renaming public interfaces.

What a good result looks like

An example, not your task: You changed code and already have an earlier review to compare with.

• agent-state.json. The revision, scoped files, finding IDs, test results, doc paths, verified status and blockers, saved for the next run.
• decision-log.md. A dated record of the branch chosen, the evidence checked and why.
• Dated state. Whether the work is new, changed, unchanged, pending or blocked.
• Branch selected and reason. Which check ran (test gaps, docs sync or bug receipt) and what made it the right one.
• Sources checked and changed findings. The files and results it looked at, and which findings changed.
• Missing facts, pending approvals and next action. What is still unknown, what needs your approval and the one step to take next.

How the work is checked

• The changed files build, and the relevant tests and linters pass. The commands and results are shown.
• Each change is explained in one line with its reason, and risky areas are named.
• Any check that could not run is reported as NOT_RUN with what it needs.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

Capabilities

• Compare saved state. Check which sources, dates or results changed since the last run.
• Choose the needed work. If the source revision is new, run the relevant test-gap and docs checks. If evidence for a finding has changed, recheck that finding and record the outcome. If a defect was repaired, prepare its receipt with VERIFIED, PARTIAL or BLOCKED based on actual checks. If the revision and evidence are unchanged, reuse the prior findings. Never equate an edited file with a passed test.
• Save the decision. Record evidence, remaining gaps and one next action. Keep external action status explicit.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id engineering-evidence-keeper only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/engineering-evidence-keeper/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• Your repository revision or commit ID
• The list of files the change touches
• Your earlier finding IDs (optional)
• Your last agent-state.json file (optional)
• Test commands and their results (optional)

Functions and deliverables

• Code change test and docs tracker: Track tests, docs and repair evidence for a code change. See what is verified, partial or still blocked.
  • A dated decision log of what was checked and why
  • An updated state file you can use on the next run
  • Test gaps, docs gaps or a bug receipt marked VERIFIED, PARTIAL or BLOCKED
  • A list of missing facts and one next action
• Software test gap finder: Share your repo or change. Get missing, weak and stale tests ranked by the behavior at risk.
  • A list of test gaps ranked from P0 (most serious) to P3.
  • For each gap, the code and test locations as path:line.
  • A specific test to add for each gap, with the scenario to check.
  • A short plan that says which tests to write first.
  • A list of checks that ran and checks that were skipped.
  Source: GitHub Copilot community.
• Code and documentation consistency checker: Share a feature or code change. Get outdated docs and missing instructions traced to the changed behavior.
  • A ranked list of stale or missing docs, from P0 to P3.
  • For each one, the code line and the doc line that disagree.
  • A plain fix direction for each finding.
  • A list of doc files most likely to need edits.
  • A note of what was checked and what was skipped.
  Source: GitHub Copilot community.
• Bug fix evidence recorder: Close a bug fix with a bug fix evidence recorder. Give it the defect and your checks. You get a receipt marked verified, partial or blocked.
  • A seven-line receipt: problem, baseline, root cause, change, proof, gaps and source
  • A status of verified, partial or blocked
  • The single next step needed if proof is missing
  • A note on whether each check was run now or supplied by you
  • A JSON version of the receipt for CI (optional, needs code running)
  Source: GitHub Copilot community.
• AI project context organiser: Share your project and next task. Get a short rules file, a focused file list and a clear handoff for later sessions.
  • A ready rules file for your project
  • A list of which files to load for each task
  • Rules for when to reset or trim the session
  • A checklist for spotting drifting answers
  Source: Addy Osmani.

Tasks you can use it for

• You changed code and already have an earlier review to compare with.
• A finding was fixed and you need proof it is closed.
• You want to know if the docs still match the code after a change.
• You are picking up work again and need to know what changed since the last run.

Limits to check before starting

• You are auditing a repo for the first time: run the test gap audit or the docs sync audit on its own, since there is no earlier review to compare against.
• Your work is not code, such as copy, ads or finance: use a tool that fits that job, because this agent only checks repository changes.
• You want a fix written for you: ask your AI to patch the bug directly, then bring the change back here to check the proof.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Open Claude, ChatGPT or Gemini and add the full skill pack, including the skills folder. Use an AI that can read and save files.
• If you have no earlier review, run the Test coverage gap investigator or the Code and docs drift checker on its own first.
• Paste your commit ID and the list of files in the change.
• Add your last agent-state.json and finding IDs if you have them.
• Say "Follow the agent. Write MISSING: for any fact you lack, and list every MISSING line in the hand-back."
• Read the hand-back. Check that each VERIFIED mark rests on a test that really ran.
• Approve any outside action yourself. It only drafts messages and never sends them.
• Next, fix what the gaps show, then run it again with the new commit ID and the saved agent-state.json.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/engineering-evidence-keeper for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.