Mentat Tasks

VerifiedSafe

Manage a local markdown-based task store with atomic claims and TTL refresh. Supports HITL/AFK workflow separation.

Sby Skills Guide Bot
ProductivityIntermediate
7/27/2026
Claude Code
#task-management#markdown#file-based#atomic-locking#multi-agent

Recommended for

Automated summary

Manages a local markdown task store (one file per task) with atomic O_EXCL sentinel claiming, TTL refresh, and HITL/AFK classification.

Strengths

  • Concrete POSIX locking protocol (O_EXCL for test-and-set, rename(2) for mutations) that prevents concurrent claims across agents
  • Clean separation of concerns: the skill owns only the schema and filesystem protocol while intake and triage are delegated
  • Explicit interface contract through a single CLI with documented subcommands and emitted events
  • Clear pick gate that forbids claiming untriaged or non-expired locked tasks

Limitations

  • Tightly coupled to an ecosystem: assumes branch conventions (holding branch, worktrees) and a shared directory via symlink
  • No typed dependency graph, write-set lease, or priority/due/estimate fields — deferred until demonstrated need
  • Requires Python3 and a hard-coded path ~/.agents/skills/mentat-tasks/scripts/tasks.py, so it is not portable as-is
When to use it

Use it when multiple agents or parallel sessions must coordinate over a local task queue without a server or database.

When not to use it

Avoid it if you want a feature-rich task manager (dependencies, priorities, estimates) or integration with external tools like GitHub Issues or Jira.

Security analysis

Safe
Quality score82/100

The allowed-tools declaration is tightly scoped to a single python3 script under the skill's own directory, and the documented operations are all local filesystem mutations (create/claim/refresh/done) using standard POSIX primitives. There are no destructive commands, no network calls, no secret/token exfiltration, no obfuscated payloads, and instructions are transparent. The only residual concern is that the actual script contents are not shown and the cross-repo symlink is a deliberate but broad filesystem action; neither rises to a destructive or exfiltrating capability.

Findings
  • •The referenced implementation (scripts/tasks.py, lib.support.frontmatter.mutate) is not included in the SKILL.md, so its actual behavior cannot be fully audited from the listing alone.
  • •Uses one-time symlink of ~/.mentat/tasks into each repo's .mentat/tasks, which shares task data across repositories (intentional, but broadens data visibility).
How this analysis is made

Examples

Créer une tâche
Crée une nouvelle tâche pour ajouter la validation d'email au formulaire d'inscription, liée au plan PRD-004, avec des critères d'acceptation en checklist.
Revendiquer et démarrer
Liste les tâches en statut todo, choisis une tâche AFK non verrouillée et revendique-la avec un TTL de 600 secondes pour l'agent claude-session-2.
Libérer une revendication expirée
La tâche T007 a un claim_expires_at dépassé ; libère-la puis revendique-la à nouveau pour l'agent claude-session-5 avec un TTL de 900 secondes.

name: mentat-tasks description: Local md task store. Atomic claim, TTL refresh, vertical-slice + HITL/AFK doctrine. allowed-tools: Bash(python3 ~/.agents/skills/mentat-tasks/scripts/tasks.py *)

mentat-tasks

Manage a local markdown task store backed by POSIX atomics. No server, no database, no binary. Each task is a file; each claim is a lock sentinel.

Layout

Tasks live at <repo>/.mentat/tasks/<ID>-{plan-ref}.md. Cross-repo unified view: ln -s ~/.mentat/tasks .mentat/tasks (user runs once).

Board-home vs worktree separation (ADR-0002): task files live on the holding branch. Chunks operating in worktrees write status back via the atomic protocol below.

Task file schema

---
id: T001
status: todo            # todo | in-progress | review | done | wontfix
kind: HITL              # HITL | AFK — vertical-slice classification (required before pick)
claimed_by: ""          # agent id or empty
claim_expires_at: ""    # RFC3339; empty if unclaimed
created_at: 2026-06-06T00:00:00Z
---

Body sections:

  • Parent — plan or PRD link.
  • What to build — one paragraph.
  • Acceptance criteria — bulleted checklist.
  • Blocked by — prose, free-form.

Invocation

All operations route through scripts/tasks.py. Set MENTAT_TASKS_DIR to override the default .mentat/tasks/ dir.

| Subcommand | Args | Description | |---|---|---| | next-id | — | Print next T### | | create | {plan-ref} | reads stdin | Create task file from stdin body | | claim <file> <agent> <ttl_s> | — | Atomic claim via O_EXCL lock sentinel | | release <file> | — | Release claim, restore todo | | refresh <file> <ttl_s> | — | Bump claim_expires_at | | done <file> | — | Terminal state: done | | wontfix <file> | — | Terminal state: wontfix | | list [--status <s>] | — | Enumerate tasks as TSV (id, status, kind, claimed_by) |

Atomic-write invariant

All frontmatter mutations use lib.support.frontmatter.mutate — tmp+os.replace same-fs rename (POSIX rename(2)). Claims use os.open(O_CREAT|O_EXCL) for exclusive create — correct POSIX primitive for test-and-set. The lock sentinel (.lock) is separate from the content file so rename never rebinds the guard inode.

Pick gate

Pick only if:

  1. status is todo, or in-progress with claim_expires_at < now (stale — release then re-claim).
  2. kind is set (HITL or AFK). Never pick untriaged.
  3. No .lock sentinel exists for a non-expired claim.

Events emitted

| Event | Payload | |---|---| | task_created | {id, slug} | | task_claimed | {id, agent, expires_at} | | task_released | {id} | | task_resolved | {id} | | task_canceled | {id} |

Boundary with other skills

Intake skill → writes task via mentat-tasks create. triage → mutates kind + status. mentat-prd → references task ids as T### in prose. mentat-tasks owns schema + filesystem protocol only.

Deferred

Typed dep graph, touches: write-set lease, priority/tags/due/estimate. Add only on demonstrated need.

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/mentat-tasks-ms2s7bye/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/mentat-tasks-ms2s7bye' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/mentat-tasks-ms2s7bye/download' --output '.skill-downloads/mentat-tasks-ms2s7bye/SKILL.md.part' && mv '.skill-downloads/mentat-tasks-ms2s7bye/SKILL.md.part' '.skill-downloads/mentat-tasks-ms2s7bye/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

Tools declared by the author (check availability in your agent): Bash(python3 ~/.agents/skills/mentat-tasks/scripts/tasks.py *)

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/mentat-tasks-ms2s7bye/SKILL.md' && test -s '.skill-downloads/mentat-tasks-ms2s7bye/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/mentat-tasks' && cp -R '.skill-downloads/mentat-tasks-ms2s7bye' '.claude/skills/mentat-tasks'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/mentat-tasks-ms2s7bye'

openskills read 'mentat-tasks-ms2s7bye'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/mentat-tasks-ms2s7bye/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills