Changelog

What's new

Every release of Wiggum CLI, from the latest to the earliest.

v0.18.3 2026-04-08 Latest 115 files +5,880 −13,287

Security hardening, npm OIDC publishing & agent scheduler

Agent backlog scheduler orchestration — automated issue-to-loop pipeline
Security hardening: CodeQL fixes, Dependabot remediation, hardened logging/workflows
npm trusted publishing via OIDC — no more manual tokens
CI stabilization and dependency bumps (actions/checkout v6, actions/setup-node v6)

v0.17.3 2026-03-18 113 files +15,831 −495

Loop optimization: --resume, merged phases & trimmed prompts

Use --resume for implementation, E2E, and review loop iterations 2+ — massive token savings
Merged verification phase into review phase to cut one full agent round
Trimmed prompt templates: removed ~160 lines of boilerplate from PROMPT_feature and PROMPT_e2e
Targeted E2E fix prompt, loop sandbox hardening, and CI check wait before merge

v0.17.2 2026-03-09 12 files +98 −6

Agent --issues flag for targeted issue processing

Added --issues flag to target specific GitHub issues by number in agent mode

v0.17.1 2026-03-09 12 files +559 −109

TUI screen improvements

Improved TUI screens — columnar issue picker, run monitor, and agent loop monitor

v0.17.0 2026-03-09 81 files +9,347 −736

Autonomous agent mode, GitHub issue integration & headless CLI

Autonomous agent mode — reads GitHub backlog, generates specs, runs loops, reviews diffs, and auto-merges PRs
GitHub issue integration for /new context phase — pull issue context directly into specs
Headless `wiggum new --auto` for non-interactive spec generation
Headless `wiggum sync` CLI command for CI pipelines
Standalone `/issue` command to browse GitHub issues from shell with columnar picker
Improved TUI screens — columnar issue picker, run monitor enhancements, agent loop monitor

v0.16.0 2026-02-23 41 files +3,542 −215

Monitor TUI routing, token tracking & CLI commands

Route `wiggum monitor` to Ink TUI RunScreen in interactive terminals
Token tracking with JSONL extraction and cache token counting
Wire run/monitor/config CLI commands with flag parsing and validation
Activity feed in RunScreen, xterm.js E2E bridge, and RunCompletionSummary layout fix

v0.15.0 2026-02-17 36 files +2,584 −125

Spec autocomplete, action inbox & AI model updates

Spec name autocomplete for /run — tab-complete spec names from your .wiggum directory
Action inbox IPC — loops can request user input mid-execution without blocking
Updated AI models to latest versions (Opus 4.6, GPT-5.1 Codex, Kimi 2.5, Gemini 3 Pro)
Polished summary Goal line with imperative sentence formatting

v0.14.0 2026-02-16 27 files +3,805 −128

Enhanced run completion summary & split review modes

Enhanced run completion summary panel with phase breakdown and timing
Split review modes: auto (review only) and merge (review + auto-merge)
SummaryBox width capped to 80 columns to prevent banner garbling
Resilient MultiSelect option parsing for AI format variations

v0.13.2 2026-02-14 2 files +6 −6

Dependency updates

Bumped qs to 6.14.2 for security patch

v0.13.1 2026-02-12 3 files +22 −26

Template fix for Claude Code

Replaced Codex CLI with Claude Code in auto review prompt template

v0.13.0 2026-02-12 32 files +2,842 −1,469

Fixed-position AppShell & background runs

Replaced Static/thread layout with fixed-position 5-zone AppShell
Background run support — Esc to background, /monitor to re-enter
Error toast notifications in AppShell
Circuit breaker for polling failures, FD leak fixes

v0.12.1 2026-02-10 7 files +14 −2,061

AppShell layout redesign & error toasts

Switched AppShell to inline layout with improved spacing
Added error toast notifications for background run failures
Fixed file descriptor leak in circuit breaker
Improved render-phase setState handling

v0.12.0 2026-02-05 4 highlights

Background runs & TUI monitor

Background run execution with real-time TUI monitoring
Loop status polling with automatic refresh
Run completion summary with phase breakdown
Header content showing active run status

v0.11.12 2026-01-28 3 highlights

Interview screen polish

Improved interview screen with better question flow
Added spec completion summary after interview
Fixed stdin handling for multi-character input

v0.11.0 2026-01-15 4 highlights

Feature loops & Claude Code support

Autonomous feature loop execution via /run command
Claude Code integration as primary coding agent
Phase-based execution: plan, implement, test, verify, PR
Configurable loop parameters and timeouts

v0.10.0 2025-12-20 4 highlights

AI-powered interviews & spec generation

Interactive AI interview for feature specification
Context-aware questions based on codebase scan
Markdown spec output compatible with any coding agent
Interview state persistence and resumption

v0.9.0 2025-12-01 4 highlights

Stack scanning & project init

Automatic tech stack detection via /init command
Framework, language, and convention mapping
Zero-config project initialization
CLAUDE.md and guide file generation