task-intelligence skill
Organize unorganized input into structured, prioritized action plans and critical paths using context from local work files.
- GitHub stars
- 37
- Files
- 3
- Last commit
- Mar 26, 2026
- License
- Apache-2.0
- Allowed tools
- —
What does the task-intelligence skill do?
task-intelligence is an agent skill. Organize unorganized input into structured, prioritized action plans and critical paths using context from local work files.
Summary generated from the skill's SKILL.md.
About the task-intelligence skill
Transforming messy brain dumps and meeting notes into structured action plans is the primary function of this skill. It is designed for professionals who need to turn unorganized thoughts, messages, or raw notes into prioritized lists with clear owners and deadlines. By reading the work.local.md file in your current directory, the agent gains the necessary context to link new tasks to existing projects and identify potential blockers.
Users can employ this tool for daily prioritization, weekly planning, or conducting backlog reviews to decide which items to keep or drop. It distinguishes between single-session items and larger projects, ensuring that complex tasks are flagged for further breakdown. The skill produces a highly structured output categorized by priority levels from P1 to P3, including a critical path and specific delegation candidates. It is not intended for basic task management or delegation tasks, which require different specialized tools.
What an agent can do with it
- 1Structure a messy brain dumpConvert raw notes and unorganized messages into a prioritized list with specific actions and project references.
- 2Determine daily work prioritiesIdentify the most important tasks for the day and generate a suggested schedule.
- 3Audit an existing backlogReview your current backlog to decide which items to keep, drop, or re-prioritize.
- 4Create integrated cross-functional plansGenerate an action plan that spans multiple departments like HR, Finance, and Operations.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the task-intelligence skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the task-intelligence skill in Plugin
/plugin marketplace add panaversity/agentfactory-business-plugins /plugin install agentic-office@agentfactory-business
Install the task-intelligence skill in Claude Code
~/.claude/skills/task-intelligence/SKILL.md
$ npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence ~/.claude/skills/task-intelligenceInstall the task-intelligence skill in Claude Code (project)
.claude/skills/task-intelligence/SKILL.md
$ npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence .claude/skills/task-intelligenceInstall the task-intelligence skill in Codex
~/.agents/skills/task-intelligence/SKILL.md
$ npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence ~/.agents/skills/task-intelligenceInstall the task-intelligence skill in Gemini CLI
~/.gemini/skills/task-intelligence/SKILL.md
$ npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence ~/.gemini/skills/task-intelligenceInstall the task-intelligence skill in Claude.ai
npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence task-intelligence && zip -r task-intelligence.zip task-intelligence # claude.ai → Customize → Skills → + → Create skill → Upload a skill → task-intelligence.zip
SKILL.md
View on GitHubTASK INTELLIGENCE WORKFLOW
STEP 1 — LOAD CONTEXT
Read work.local.md in the current working directory.
If it does not exist, tell the user to run /agentic-office:setup first.
STEP 2 — IDENTIFY REQUEST TYPE
TYPE 1: CAPTURE (from brain dump / meeting notes / messages) Purpose: Rapidly capture and structure unorganised input. Input: Raw notes, brain dump, meeting output, messages — any format Output: Structured, prioritised list with owners and deadlines
TYPE 2: DAILY PRIORITISATION Purpose: Determine what to work on today from a larger set. Input: Full list (from work.local.md or provided) Output: Today's critical path (3-5 items) + schedule suggestion
TYPE 3: WEEKLY PLANNING Purpose: Set the week's priorities before it begins. Input: Project statuses; open delegations; upcoming deadlines Output: Week-ahead critical path; delegation candidates; decision items
TYPE 4: BACKLOG REVIEW Purpose: Audit the backlog — drop dead items; re-prioritise the rest. Input: Current backlog Output: Kept / dropped / re-prioritised list with rationale
TYPE 5: CROSS-DOMAIN PLAN Purpose: Generate an integrated action plan for a scenario that spans multiple functions (HR + Finance + Ops, etc.) Output: Items by function; dependencies; delegations; timeline
STEP 3 — EXECUTE WITH PRIORITY SORTING
OUTPUT STRUCTURE:
CAPTURE — [Date]
================================================================
[N] items captured | [N] overdue | [N] delegation candidates
-- P1 — URGENT / HIGH IMPACT ---------------------------------
[ ] [ITEM-NNN] [Title]
Context: [Why this matters; what it connects to]
Due: [Specific date or "today"]
Project: [Project name from work.local.md]
Action: [Specific first action — not vague]
Delegate: [If applicable — to whom]
Note: [Any sensitivity or communication guidance]
-- P2 — IMPORTANT / THIS WEEK --------------------------------
[Same structure]
-- P3 — STANDARD / BACKLOG -----------------------------------
[Same structure — briefer]
-- DELEGATION CANDIDATES --------------------------------------
[Item] → [Person from work.local.md] (with brief note on approach)
-- BLOCKED / AT RISK ------------------------------------------
[Any item that cannot proceed without something else; name the blocker]
================================================================
Priority Sorting Logic
When determining priority, apply in order:
- Is there a hard deadline? (Date-certain = prioritise)
- Is it blocking someone else? (Blocking = elevate)
- Does it affect a P1 project? (P1 project item = at least P2)
- What is the consequence if it slips one week? (Assess impact honestly)
- Is the urgency feeling real or just habitual? (Email = rarely P1)
The Brain Dump Pattern
When user provides unorganised input:
- Extract every distinct item (even if vague or incomplete)
- Enrich each with project context from work.local.md
- Identify dependencies between items
- Flag anything that should be delegated
- Flag anything that should be dropped or questioned
Item vs. Project Distinction
ITEM: Can be completed in one session; has a single clear output PROJECT: Multiple items; spans days or weeks; has milestones
If an "item" has >3 steps and spans >1 day: it is a project.
Create a project entry in work.local.md instead.
Breaking projects down into single-session items is the user's job —
or use /agentic-office:progress-tracker to create a milestone plan.
NEVER DO THESE
- NEVER classify more than 5 items as P1 — force re-prioritisation if the user tries to make everything critical
- NEVER leave a captured item without a project reference if it connects to a known project in work.local.md
- NEVER produce a list without a critical path — a list of everything is not a plan; a critical path is
- NEVER mark an item as P1 based purely on urgency without assessing actual importance
- NEVER produce a weekly plan without surfacing open delegations
Files in this skill
3task-intelligence skill questions
How do I install the task-intelligence skill in Claude Code?
Run npx degit panaversity/agentfactory-business-plugins/agentic-office/skills/task-intelligence ~/.claude/skills/task-intelligence. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the task-intelligence skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
What should I do if the work.local.md file is missing?
You should run the /agentic-office:setup command first to create the necessary context.
How does the skill distinguish between an item and a project?
An item is something that can be completed in one session with a single output, whereas a project spans multiple days or weeks and has milestones. If an item has more than three steps and spans more than one day, it should be treated as a project.
Can I use this skill for basic task creation and deletion?
No, this skill is not for basic task CRUD; you should use the official productivity plugin task-management for those actions.