- Home
- Skills
- Agents & Orchestration
- workmap - project work map for AI agents
More screenshots
Works with the AI Agents you already use
workmap - project work map for AI agents
"Where are we on the project?" Open the map instead of asking the agent.
$19
workmap - project work map for AI agents
Example session with this skill installed
Set up a work map for this repo (Acme Commerce), then log today's work: OAuth sign-in is shipped, session token refresh is in progress by backend-agent, and we dropped magic-link sign-in. Show me where the project stands.
- Read your context and instructions
- Compiled the workmap - project
- Generated the document
Installed the work map and logged today's work.
Install
bash scripts/init.sh . --name "Acme Commerce"
Created scripts/workmap.py, scripts/workmap.template.html, docs/shared/workmap.md → docs/workmap.html; appended the work-map rules and the reporting protocol to CLAUDE.md.
Logged
python3 scripts/workmap.py add "Platform & Core API/Authentication Service" "OAuth sign-in flow" -s done -o backend-agent -n "Google and GitHub providers"
# added: T1 · OAuth sign-in flow (done) @backend-agent under Platform & Core API/Authentication Service
python3 scripts/workmap.py add P1 "Session token refresh" -s doing -o backend-agent -n "rotating refresh tokens"
# added: T2 · Session token refresh (in progress) @backend-agent under Platform & Core API/Authentication Service
python3 scripts/workmap.py add P1 "Magic link sign-in" -s cancel -n "dropped: OAuth covers the need"
# added: T3 · Magic link sign-in (cancelled) under Platform & Core API/Authentication Service
Where the project stands (python3 scripts/workmap.py show -d 3)
▶ Acme Commerce
▶ 1 Platform & Core API [epic]
▶ P1 Authentication Service [feature]
✓ T1 OAuth sign-in flow @backend-agent — Google and GitHub providers
▶ T2 Session token refresh @backend-agent — rotating refresh tokens
✕ T3 Magic link sign-in — dropped: OAuth covers the need
✓ 1 done · ▶ 1 in progress · ○ 0 pending · ‖ 0 paused · ✕ 1 cancelled · 50% done (1/2)
Open docs/workmap.html for the map, or run python3 -m http.server 4590 --bind 127.0.0.1 from the project root and open http://127.0.0.1:4590/docs/.
Workmap report
- changed: T1 OAuth sign-in flow → done · T2 Session token refresh → doing · T3 Magic link sign-in → cancel
- in progress: T2 Session token refresh @backend-agent
- map: docs/workmap.html
Next steps
- Add the remaining Epics and Features with
add "<Epic>/<Feature>" "<Work item>"; 5-9 Epics is a good size. - Every agent that loads the skill now follows the reporting protocol; paste
references/reporting.mdinto subagent briefs so they report back their node id, owner, status and result.
template.html
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
"Where are we on the project?" Open the map instead of asking the agent.
Coding agents forget what they did last session, and you cannot see what they are doing now. workmap gives the project one page that the agents themselves keep current: docs/workmap.html, every Epic, Feature, Task and Subtask with a stable id (1, P1, T1, B1), a status (pending, in progress, paused, cancelled, done) and the name of the agent or person on it. A progress bar at the top says how far the project is.
The agents keep the map with four commands, show, find, add and set, a few tokens each. Every change lands in docs/shared/workmap.md, a plain markdown outline: grep it, diff it, read it in any editor. Nothing hides in a database.
What makes it hold is the reporting protocol that loads with the skill. Before a piece of work starts, the agent logs it. When it finishes, the agent marks it. When it delegates, the subagent gets the same rules and reports back. Every turn ends with a Workmap report. Done means the map says done.
What's included
- Work map page: status-coloured chips, a running light on work in progress, owner tags, expand/collapse, full screen, light and dark theme, phone-safe. One self-contained file that opens from
file://. - CLI
scripts/workmap.py: stable ids, computed parent status, duplicate protection, owners, task-file integration (docs/tasks/T-*.md), naming nudges. - Markdown source
docs/shared/workmap.md; an older JSON map converts automatically. - Mandatory reporting protocol, a paste-ready block for subagent briefs, and project rules written into
CLAUDE.mdorAGENTS.md. - Pairs with the free workdocs skill (same author): six documentation pages with interactive architecture, workflow, lifecycle and data-flow diagrams, in the same look; the map page picks up their navigation and shares the light / dark choice, so the two read as one site.
- English and Vietnamese labels.
How it works
bash scripts/init.sh <project> --name "<Project>"installs the map: one script, one template, one markdown file, one self-contained page.- Agents log work with
add "<Epic>/<Feature>" "<Work item>" -s doing -o <agent>and finish it withset <id> done -n "<result>". show -d 2answers "where are we?" in a few tokens;docs/workmap.htmlshows it to everyone.
Why this beats a prompt
A prompt gives you one session's good intentions. workmap gives every agent the same rules, one file they all read and write, ids that never change, and a report at the end of every turn. The map is the audit trail.
Requirements
python3 3.8+ and bash (macOS, Linux, WSL). No packages, no network access, no elevated permissions. Works with Claude Code and any agent that supports the SKILL.md standard.
Privacy
No telemetry, no network calls. The installer creates files inside your project, never overwrites, and only appends clearly marked blocks to README.md and the agent rules file.
Known limitations
Node names should stay between 12 and 26 characters so the chips line up (the CLI nudges you). Owners are single tokens (backend-agent, linh).
Licence and support
Files installed into your project are yours to use, modify and ship, commercial or open source; the skill itself may not be redistributed. Support: nnquangit@gmail.com
Requires python3 3.8+ and bash; no network access, no packages. Node names should be 12-26 characters so the chips line up (the CLI nudges you). Owners are single tokens such as backend-agent. The optional docs-kit pages fetch your markdown, so they need a local static server; the work map page itself opens from file://.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 10 days ago
Requires