Install
$ agentstack add skill-avanrossum-claude-task-skills-task-audit ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
/task-audit — Progressive Task Audit
You are walking the user through a structured audit of all open tasks. This is a collaborative, progressive process — not a dump of information.
Arguments: $ARGUMENTS
Philosophy
A task audit is a conversation, not a report. The goal is to surface stale tasks, update governance, resolve ambiguity, and help the user decide what to prioritize. You lead the process; the user makes decisions.
Step 1 — Silent Governance Pre-Pass
Before involving the user, do a quick scan of all open tasks silently:
- Find the task tracker (
TASKS.mdor equivalent) and read it - For each open/in-progress task, read its task document (TASK.md or equivalent)
- Note any issues you find — but don't fix them yet. Flag them for discussion:
- Stale status — task says "In Progress" but no changes in weeks
- Missing governance — task folder exists but TASK.md is incomplete or missing
- Orphaned entries — tracker references a task that doesn't exist, or a folder exists without a tracker entry
- Ambiguous state — status is unclear, or the task document contradicts the tracker
- Scope creep — spec says one thing but changes made suggest a different scope
- Unresolved blockers — task says "blocked" with no plan to unblock
- Governance debt — decisions made but not documented, open questions with no owner
Step 2 — Opening Summary
Present a brief overview to the user:
Found N open tasks. Quick health check:
- X tasks look healthy (governance current, status accurate)
- Y tasks have issues I want to walk through with you
- Z tasks may be stale or ready to close
Then ask: "Ready to walk through them? I'll go one at a time, starting with the ones that need attention."
Step 3 — Progressive Walk-Through
Go through tasks one at a time, starting with the ones that have issues (from Step 1). For each task:
Present
- Task name and current status
- One-line summary of what it is
- Any issues found in the pre-pass
Ask
Depending on the issues, ask the relevant questions:
- "This has been in progress since [date] with no recent updates. Is it still active, or should we mark it stalled/closed?"
- "The spec says X but the changes suggest Y. Which is accurate?"
- "This is marked blocked. What's the blocker and is there a path to unblock?"
- "This looks done. Should we close it?"
- "The task document is missing [section]. Want me to fill it in based on what I can see?"
Act
After the user responds:
- Update the task document immediately
- Update the task tracker if status changed
- Note any follow-up items
Then move to the next task. Wait for the user's response before proceeding — don't batch multiple tasks into one message.
Step 4 — Healthy Tasks (Quick Confirmation)
After addressing problem tasks, briefly run through the healthy ones:
"These N tasks look current and healthy — just confirming:"
- Task A — [status] — still accurate?
- Task B — [status] — still accurate?
The user can confirm all at once or flag any that need discussion.
Step 5 — Wrap-Up
After all tasks are reviewed:
- Summarize what changed:
- Tasks closed: N
- Tasks updated: N
- New blockers identified: N
- Tasks still healthy: N
- Update the task tracker to reflect all changes made during the audit
- If appropriate, suggest priority order for remaining open tasks
Keep the wrap-up brief. The value was in the conversation, not the summary.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: avanrossum
- Source: avanrossum/claude-task-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.