Td Task Management

What it does

When your agent uses it

• Agents need to track work
• Maintain context across sessions

Example prompts

• “/td-task-management”

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

1. 1Single focused issue → Use single-issue workflow
2. 2Multiple related issues → Use td ws start for work sessions
3. 3Work will continue elsewhere → Use td handoff or td ws handoff
4. 4A decision aids continuity → Log it with --decision
5. 5An uncertainty matters → Record it with --uncertain
6. 6Track files → Use td link so future sessions know what changed

What it can do on your machine

Read from SKILL.md and the folder at commit b206bbf. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  No scripts in the folder and no shell commands in SKILL.md (its code samples are bash).

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Td Task Management loads about 2k tokens when it runs, and up to ~3.5k if it reads all its reference files. Until then it costs about 89 tokens; SKILL.md has 720 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~89

When it runs · the whole SKILL.md, loaded when a task matches

~2k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~3.5k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from marcus/td at commit b206bbf, republished under its MIT licence (© marcus). 720 words, ~1,993 tokens.

name

td-task-management

description

Task management for AI agents across context windows. Use when agents need to track work, log progress, hand off state, and maintain context across sessions. Includes workflows for single-issue focus, multi-issue work sessions, and structured handoffs. Essential for AI-assisted development where context windows reset between sessions.

td usage --new-session -q  # Auto-rotate and show compact current state

td start <issue-id>                    # Begin work
td log "OAuth callback implemented"    # Track progress
td log --decision "Using JWT tokens"   # Log decisions
td handoff <id> --done "..." --remaining "..."  # Capture state
td review <id>                         # Submit for review

td ws start "Auth implementation"      # Start work session
td ws tag td-a1b2 td-c3d4             # Associate issues (auto-starts them)
td ws tag --no-start td-e5f6          # Associate without starting
td ws log "Shared token storage"       # Log to all tagged issues
td ws handoff                          # Capture state, end session

# 1. Check what to work on
td usage          # See current state
td next           # Highest priority open issue
td critical-path  # What unblocks most work

# 2. Start work
td start <id>

# 3. Begin logging
td log "Started implementation"

td handoff <id> \
  --done "OAuth flow, token storage" \
  --remaining "Refresh token rotation, error handling" \
  --decision "Using JWT for stateless auth" \
  --uncertain "Should tokens expire on password change?"

# 1. See reviewable issues
td reviewable

# 2. Check details
td show <id>
td context <id>

# 3. Approve or reject
# Independent review:
td approve <id> --reason "Reviewed diff, looks good"
# Or, in trusted mode, when you implemented it yourself:
#   a sub-agent reviewed it — name them (no --reason needed):
td approve <id> --reviewed-by "code-reviewer sub-agent"
#   you reviewed it yourself — say so:
td approve <id> --self-review --reason "Reviewed own diff, tests pass"
# Or:
td reject <id> --reason "Missing error handling"

# 1. Log the blocker
td log --blocker "Waiting on API spec from backend team"

# 2. Work on something else
td next              # Get another issue
td ws tag td-e5f6   # Add to work session

# 3. Come back to blocked issue later
td context td-a1b2  # Refresh context when blocker resolves

open → in_progress → in_review → closed
         |              |
         v              | (reject)
     blocked -----------+

td usage --new-session -q

© marcus, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 2 other files (references) in td-task-management of marcus/td.

• SKILL.md
• references/ai_agent_workflows.md
• references/quick_reference.md

Open the folder on GitHubat commit b206bbf