issue-assistant 0.2.2

Deterministic-first GitHub issue triage engine

Issue Assistant

Issue Assistant is a deterministic GitHub issue triage engine designed for maintainers who want consistent, explainable signals without automation they can’t trust.

It analyzes GitHub issues and produces Markdown and JSON artifacts that help answer:

• Is this issue actionable or missing critical information?
• Does it appear to be a duplicate of existing work?
• How much maintainer effort is likely required?
• Why did the tool reach this conclusion?

All outputs are written to .issue-assistant/. Nothing is hidden, guessed, or implicit.

---

Why this exists

Maintainers spend a disproportionate amount of time triaging issues that are:

• missing reproduction steps, logs, or environment details
• duplicates that fragment discussion
• unclear about expected vs actual behavior
• noisy or low-signal

Issue Assistant provides a deterministic baseline for triage so humans can make faster, more consistent decisions without surrendering control.

---

Design principles

Issue Assistant is intentionally conservative:

• Deterministic only — no LLMs, no probabilistic output
• Explainable by construction — every outcome maps to rules and evidence
• Governance-aware — safe defaults, no surprise side effects
• Human-first — maintainers always make the final call

If a signal cannot be explained clearly, it does not exist.

---

Installation

Requires Python 3.10+.

pip install issue-assistant

---

Run locally

issue-assistant analyze \
  --github-token "$GITHUB_TOKEN" \
  --repo owner/name \
  --output-dir .issue-assistant \
  --governance-mode dry-run

This performs a full analysis and writes artifacts locally without modifying issues.

---

GitHub Action

Issue Assistant is available as a GitHub Action:

uses: siggmond/issue-assistant@v0.2.0

The Action is a thin wrapper around the CLI. You can use the CLI independently or rely entirely on the Action.

Typical triggers include:

• issue opened / edited / reopened
• scheduled runs (weekly digest)

Artifacts can be uploaded or committed depending on your workflow.

---

Governance modes

Mode	Comments	Label suggestions	Intended use
dry-run	never	never	analysis only; safest default
strict	limited	no	request missing info, flag duplicates
aggressive	yes	yes (suggestions)	higher throughput without auto-closing

Hard constraints:

• issues are never auto-closed
• fixes are never guessed
• maintainer decisions are never overridden

---

Artifacts generated

A typical run produces:

• ISSUE_SUMMARY.md / issues.json
• MAINTAINER_LOAD.md / maintainer_load.json
• EXPLAINABILITY.md / explainability.json
• per-issue playbooks and explainability JSON

Example outputs are included under docs/screenshots/.

---

When not to use this tool

Issue Assistant is not a good fit if you want:

• fully automated triage or auto-closing
• subjective or sentiment-based analysis
• invisible automation
• zero repository artifacts

---

Philosophy

Explainability is a core invariant.

If the tool cannot explain why it produced an output, that output should not exist.