laird/agents

laird/agents

Command

Approve one or more AI-generated proposals for implementation by removing the proposal label.

laird/agents

Command

Start by examining the environment to give contextual guidance.

laird/agents

Command

Use AI-powered brainstorming to explore design options, clarify requirements, and develop implementation approaches for a GitHub issue.

laird/agents

Command

Resolve and close an issue with an optional closing comment.

laird/agents

Command

Phase 3 fix-loop token-efficiency (spec §5.3, §10 Phase 3, §7.1.1 verified 2026-05-21). Haiku gate; fix work dispatched to Sonnet/Opus via Task model=.

laird/agents

Command

Wrapper around /autocoder:fix that runs it in a loop forever.

laird/agents

Command

Analyze all open GitHub issues, prioritize them, and begin systematically fixing or implementing them starting with the highest priority.

laird/agents

Command

Run the complete regression test suite, analyze all results (successes and failures), and manage GitHub issues accordingly.

laird/agents

Command

Slim fix-loop gate. Per spec §5.3 / §7.1.1, no model: frontmatter: inherits the model from whatever invokes it (fix-loop's loop model). A Haiku-pinned variant (/autocoder:dispatch) is Phase 3, not here. -->.

laird/agents

Command

Improve test coverage using persistent coverage report.

laird/agents

Command

Complete installation of autocoder plugin components for autonomous issue resolution and parallel agent coordination.

laird/agents

Command

List open issues, optionally filtered by label or priority.

laird/agents

Command

Display all open issues that require design work before implementation. Uses the configured issue source (file or GitHub).

laird/agents

Command

Display all open issues that require human feedback or clarification. Uses the configured issue source (file or GitHub).

laird/agents

Command

Display all AI-generated enhancement proposals awaiting human review and approval. Uses the configured issue source (file or GitHub).

laird/agents

Command

Snapshot manager session state to MANAGER-STATE.md in the project root, then guide the manager through a clean context reset. Run /autocoder:manager-resume in the fresh session to reload.

laird/agents

Command

Load a fresh manager session with full situational awareness. Reads MANAGER-STATE.md (saved by /autocoder:manager-handoff), then queries live GitHub state and worker screens to show the delta since the save.

laird/agents

Command

One-shot migration from the legacy flat .issues/.md layout to the four-bucket layout (open/, working/, blocked/, closed/). Idempotent — refuses to run if any bucket already contains .md files.

laird/agents

Command

Run /monitor-workers in a continuous loop. This is the default command for the manager session in a parallel agent swarm.

laird/agents

Command

Monitor worker agents in worktrees, detect stale work, assign unblocked issues to idle workers via cmux/tmux, and deploy when all work is complete.

laird/agents

Command

Create a new issue in the configured issue backend (GitHub Issues or file backend).

laird/agents

Command

Analyze accumulated project history to produce 3–5 specific, evidence-backed recommendations for improving the autocoder workflow. Output: IMPROVEMENTS.md.

laird/agents

Command

Interactive planning session to review and unblock issues that fix-loop has labeled with blocking tags. This command is designed to run in parallel with /fix-loop in a separate terminal session.

laird/agents

Command

Switch the configured issue backend. Optionally migrates existing issues.