// File and manage megaplan tickets — short, repo-scoped notes on problems or observations that get folded into epics and auto-addressed when the resolving epic completes.
Pick the right megaplan profile, thinking-strength tier, and robustness level for the work in front of you — for both Codex and Claude harnesses. Consult before invoking megaplan.
Three-round adversarial critique of epic drafts (high / mid / low abstraction) with revision after each round. Produces a chain-ready revised epic.
Observe an in-flight megaplan — introspect state, trace events, diagnose blockages, detect drift. Companion to megaplan-decision. Use during and after a run, not before.
AI agent harness for coordinating Claude and GPT to make and execute extremely robust plans.
Methodology for running multi-profile LLM bake-offs via megaplan and presenting fair, blind-assessed comparisons. Cost/quality discipline, prompt hygiene, pre-merge gates, and reporting patterns. Use when the user says "bakeoff", "bake off", "megaplan bakeoff", or asks to compare profile mixes head-to-head.
Run megaplan plans and chains inside a provider-managed container (today, Railway) with a persistent workspace volume. Use when the run needs to outlast a local terminal session, span multiple repos, or share a long-lived dev box across concurrent chains. Covers `cloud.yaml` fields, `extra_repos[]` + `chain_session` multi-tenancy, the operator loop, and the gotchas that wedge fresh runs.
| name | megaplan-tickets |
| description | File and manage megaplan tickets — short, repo-scoped notes on problems or observations that get folded into epics and auto-addressed when the resolving epic completes. |
Tickets are short, repo-scoped notes on problems, bugs, or "we should look at this" observations against a codebase. They live as committed .md files under <repo>/.megaplan/tickets/ and (when a cloud store is configured) mirror to tickets / ticket_epics tables. Tickets are inert until folded into an epic — they capture problems, not work.
Reach for megaplan ticket new when:
Do not use tickets for:
megaplan feedback (a separate, plan-scoped concept).megaplan ticket new "<title>" -b "<body>" [--tags tag1,tag2]
Conventions:
-b "...", --edit (open $EDITOR), and - (read body from stdin) are mutually exclusive; one is required.ticket new prints only the ULID on stdout on success (logs go to stderr). Capture it for piping.--tags is comma-separated. Tags are freeform; common ones: bug, refactor, tech-debt, observability, docs, cross-repo.source is auto-derived: if MEGAPLAN_TURN_ID is set (the agent is running under a megaplan-launched worker), source=agent and the turn id is recorded; otherwise source=human.git rev-list --max-parents=0 HEAD). Identity survives rename / transfer / remote change.Multi-line body via stdin:
cat <<'EOF' | megaplan ticket new "Title here" --tags tag1,tag2 -
First paragraph of the body.
More detail in a second paragraph.
EOF
megaplan ticket list [--status open|addressed|dismissed] [--tags t1,t2] [--json]
megaplan ticket show <id> [--json]
megaplan ticket search [KW ...] [--all] [--project P]... [--all-projects]
[--status ...] [--tags ...]
[--sort created|edited|length|title] [--asc]
[--limit N] [--json] [--no-snippet]
--json on read commands emits structured output; safe for piping.
ticket search is the cross-cutting reader. Defaults:
--all-projects to scan every known repo (locally: the auto-maintained ~/.config/megaplan/known_repos.json registry; cloud: every codebase). Pass --project PATH|owner/name|name (repeatable) to scope to specific repos.--all to require all keywords. Match is case-insensitive substring across title, body, tags, and resolution_note.--sort {created,edited,length,title}; --asc flips to ascending. Default: created, descending.--no-snippet to hide.Examples:
# Anything mentioning "stderr" or "timeout" in this repo:
megaplan ticket search stderr timeout
# Same, but require BOTH terms:
megaplan ticket search stderr timeout --all
# Across every repo on this machine, only open, sorted by longest body first:
megaplan ticket search redis --all-projects --status open --sort length --json
# Scoped to two specific repos:
megaplan ticket search auth --project ~/Documents/reigh-app --project banodoco/megaplan
megaplan ticket edit <id> [--title ...] [--body ...] [--status ...] [--add-tag ...] [--remove-tag ...]
megaplan ticket link <ticket> <epic> [--resolves]
megaplan ticket unlink <ticket> <epic>
megaplan ticket addressed <id> [--note "..."]
megaplan ticket dismiss <id> [--reason "..."]
megaplan ticket reopen <id>
link ... --resolves marks the join with resolves_on_complete=true. When the linked epic transitions to done, the ticket auto-flips to addressed and gets a resolution_note referencing the epic. Idempotent — re-running on an already-addressed ticket is a no-op.
Megaplan runs in one of two modes; tickets work transparently in both:
.md files on disk are the sole source of truth. codebase_id will be null in frontmatter; identity is computed on demand from the repo's root SHA.SUPABASE_DB_URL or megaplan config points at a store) — every operation writes the .md file and mirrors to the tickets / ticket_epics tables. Auto-registers the codebase on first ticket if needed.You do not need to detect the mode yourself; the CLI dispatches based on whether a store is configured.
<repo>/.megaplan/tickets/{ulid}-{slug}.md
---
id: 01HXY...
title: Execute step retries swallow stderr on timeout
tags: [execute, observability]
status: open
source: agent
codebase_id: null # null in local-only, populated in cloud
created_at: 2026-05-11T14:22:00Z
epics:
- id: epic_01HZ...
resolves_on_complete: true
---
# Body
Markdown prose describing the issue.
Commit ticket files. The .md is the human-readable artifact and (in local-only) the system of record.
When a new epic is created or refined for the current repo, the planner automatically surfaces open tickets ranked by tag overlap with the epic's goal and recency. The planner may propose links with resolves_on_complete=true. You don't need to remind it — discovery is built into the plan-phase prompt assembly.
| You want to … | Run |
|---|---|
| File a ticket | megaplan ticket new "title" -b "body" [--tags t1,t2] |
| Pipe a multi-line body | cat body.md | megaplan ticket new "title" - |
| List open tickets | megaplan ticket list --status open --json |
| Search by keyword in this repo | megaplan ticket search foo bar |
| Search across every repo | megaplan ticket search foo --all-projects |
| Link a ticket to an epic so it auto-closes | megaplan ticket link <tid> <eid> --resolves |
| Mark addressed manually | megaplan ticket addressed <tid> --note "..." |
| Reopen a closed ticket | megaplan ticket reopen <tid> |
That's the whole surface. The auto-address hook does the rest.