todo-tracker

Tracks durable work items across workstations and agents using git-backed context packs.

Updated Sep 23, 2026
One-click install
npx skills add https://github.com/drgaciw/rlm-base-dev-fork --skill todo-tracker-drgaciw
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: todo-tracker
Source: https://github.com/drgaciw/rlm-base-dev-fork/tree/main/.cursor/skills/todo-tracker
Command: npx skills add https://github.com/drgaciw/rlm-base-dev-fork --skill todo-tracker-drgaciw

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Session-based task lists disappear when a session ends, and parallel work across machines or agents silently duplicates effort. This Skill provides a durable, git-backed tracker for finding, claiming, and closing work items so ownership and progress survive across workstations, agents, and review rounds. ## Core Features & Use Cases - Durable context packs: Each work item is a pack containing the goal, dated evidence, rejected approaches, and verification steps, so any agent can start cold without re-deriving prior work. - Claim and closeout protocol: index.py claim|release|close stamps, validates, commits, and pushes in one step, making simultaneous starts collide instead of duplicating work. - Sequencing and validation: blocked_by/blocks and bundle frontmatter fields express ordering and batching, while --check validates every pack and detects stale indexes or closeout violations. - Use Case: Starting a session with no idea what is in flight, run index.py list --status open, read a pack with show, claim it, and close it out after acceptance gates are met. ## Quick Start Ask the agent to list open work items from the tracker index, show the full pack for one item, and claim it before starting work.

Frequently Asked Questions about todo-tracker

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I track work items across multiple machines and agents?▼

Store each work item as a context pack in a git-cloned artifacts repository, then use index.py to list, claim, and close packs. The push on claim makes simultaneous starts collide instead of silently duplicating work.

How do I claim a work item before starting work?▼

Run `python .agents/artifacts/todos/index.py claim <id>`, which pulls, stamps, commits, and pushes in one step. It refuses an already-claimed pack and reports who holds it and since when.

Can I use this tracker without access to the private artifacts repository?▼

No, the tracker requires cloning the private rlm-base-artifacts repository to .agents/artifacts/. If the clone fails, continue without the tracker; it is optional and not required for using the public skills library.

Why does index.py --check exit 1 even when nothing is wrong?▼

The index embeds computed claim ages, so time alone stales it and a stale index exits 1 exactly like a real closeout violation. Read the message: 'INDEX.md is stale' means regenerate, while 'Closeout problems' means fix the pack.

When should I release a claimed work item?▼

Release only when actually handing off, abandoning, or explicitly parking the item, and record the reason in the pack log. Ending a turn, finishing a review round, or waiting for merge approval is not a release event.