kiloloop

Run the author side of the review loop: request review, poll for feedback, optionally delegate triage or isolated fixes to Codex subagents, address findings, and report progress through inbox plus PR comments.

Run the author side of the review loop — address findings and drive to LGTM. Single-pass by default (exits after sending messages); use --poll for in-session polling. Hybrid split — leader handles Bash I/O, subagent does code fixes only.

Run the reviewer side of the review loop: gather PR context, optionally delegate read-only review analysis to a Codex subagent, write findings packets, and send feedback or LGTM.

Run the reviewer side of the review loop — review PR diffs, produce findings, and evaluate quality gate. Single-pass by default (exits after verdict); use --poll for in-session round 2 polling. Hybrid split — leader handles Bash I/O, subagent does analysis only.

Synthesize org-memory — fold new events into recent.md, promote cross-repo patterns to decisions.md and rules.md, and audit for staleness, cross-file conflicts, and drift. Invoke whenever the user asks "synthesize org-memory", "fold events", "is recent.md stale", "audit org-memory", or any phrasing like "did we synthesize org-memory" — even if they do not explicitly type the slash command.

Synthesize org-memory — fold new events into recent.md, promote cross-repo patterns to decisions.md and rules.md, and audit for staleness, cross-file conflicts, and drift. Invoke whenever the user asks "synthesize org-memory", "fold events", "is recent.md stale", "audit org-memory", or any phrasing like "did we synthesize org-memory" — even if they don't explicitly type the slash command.

Process the current Codex project inbox in a single pass. Use when triaging pending inbox YAML messages or handling notification/question/review-loop traffic through the OACP CLI. Honors OACP Phase 1 receiver autonomy (`always_pause` / `auto_review`) per the active OACP autonomy protocol.

Check a project's OACP inbox for new agent messages and auto-act on them by message type. Single-pass processor — pair with an event-driven `oacp watch` runner (preferred) or the `/loop 2m /check-inbox` fallback for continuous monitoring. Honors OACP Phase 1 receiver autonomy (`always_pause` / `auto_review`) with a 4-gate evaluator, audit events, and a threshold-exceeded checkpoint.

Vault read/write primitives for an optional Obsidian vault. Handles daily-log writes, decision filing, weekly reviews, reading-list maintenance, archiving, and lint. Use this whenever the user types /obsidian, wants to file a decision into their vault, asks to log something to today's daily note, mentions "vault" or "Obsidian," needs vault hygiene checks, or another skill (morning, evening, housekeep) needs to read or write the vault. No-op silently if config.yaml has no vault_dir.

Audits "Learned from runs" sections across .claude/skills/ — promotes durable lessons into the main SKILL.md body, prunes stale or already-covered entries. Use this whenever the user types /consolidate-learning, /housekeep flags a skill with a bloated learned section, the user asks to clean up skill learnings, says "this skill is getting cluttered," or a monthly cadence cycle hits. Conservative — proposes diffs for user approval before applying.

Weekly hygiene pass over Iantha's memory, skills, git state, and (optional) vault. Flags stale tasks, drift in memory file timestamps, uncommitted changes, vault `_to-delete/` clutter, and skills with bloated "Learned from runs" sections. Use this whenever the user types /housekeep, mentions cleaning up / "things feel cluttered" / "let's tidy up", asks for a weekly check-in, or it's been a week since the last housekeep commit.

Daily start-of-day briefing — reads memory, surfaces today's tasks, overdue items, recurring obligations, and personal-life reminders. Use this whenever the user types /morning, asks "what's on my plate today" / "what should I focus on" / "kick off the day", starts a new day's session and hasn't been briefed yet, or wants a daily standup-style rundown of personal life. If a vault is configured, can also write the briefing to today's daily note.

First-run onboarding interview — offers to seed the blank memory files with the user's real life through a short, skippable Q&A, writes the answers into the live `.md` files in canonical format, then runs /morning on the real data. Use this whenever the user types /setup or /first-run, on a fresh clone with blank memory when they want to get started instead of learning-as-you-talk, or later to gap-fill memory sections that are still empty. Offered, never forced — a decline keeps the default learn-as-you-go behavior.

Capture decisions, learnings, and feedback from the current session into memory/{decisions,learnings,feedback}.md. Use this whenever the user types /debrief, says "let's note what we decided" / "capture this" / "save this for next time", finishes a substantive session worth remembering across sessions, or wants to file a discrete decision with rationale. Single-agent personal version (vs cortex /debrief which is OACP-multi-agent).

End-of-day wrap-up — archives completed tasks from memory, rolls over recurring obligations, surfaces tomorrow's items, and optionally writes a daily log to the user's Obsidian vault. Use this whenever the user types /evening, says "wrap up the day" / "close out" / "let's call it", finishes a substantive day of work, or wants to commit memory updates before stopping.

Capture a structured session summary and send it to the cortex inbox. Run at the end of any Claude Code session.

Consolidate session debriefs into a single source of truth (SSOT). Run once daily from the cortex repo.

Capture a structured session summary and send it to the cortex OACP workspace. Run at the end of any Codex session.

Consolidate session debriefs into a single source of truth (SSOT). Run once daily from the cortex repo.

Estimate effort for AI coding agent tasks using PERT three-point estimation with METR reliability thresholds and wave planning.

Codex skill for running agent-estimate CLI commands (estimate, validate, calibrate).