AgentStack
SKILL verified MIT Self-run

Agent Task Shaping

skill-45ck-skill-harness-agent-task-shaping · by 45ck

Shape ambiguous or frontier-agent work into a bounded agent task with outcome, scope, context, tools, autonomy level, and verification evidence.

No reviews yet
0 installs
13 views
0.0% view→install

Install

$ agentstack add skill-45ck-skill-harness-agent-task-shaping

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No 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.

Are you the author of Agent Task Shaping? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Agent Task Shaping

Use this skill before launching an agentic implementation, research, review, or long-running workflow from a vague request.

Process

  • State the user-visible outcome in one sentence.
  • Name the smallest reversible slice that can produce evidence.
  • Separate required context from optional background.
  • Choose the agent loadout or role that can own the slice end to end.
  • Identify allowed tools, blocked tools, and any permission expansion that needs approval.
  • Define the autonomy level: act directly, ask before mutation, ask before external side effects, or propose only.
  • Define verification evidence before work starts.

Output

Task

Scope

Context Needed

Agent / Loadout

Tool Boundaries

Autonomy Level

Verification Evidence

Follow-Up Capture

Avoid

  • turning broad strategy into unbounded execution
  • starting from tool availability instead of task outcome
  • asking a human for decisions the agent can resolve from repository evidence
  • adopting learnings without a traceable issue, test, policy check, or run receipt

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

  • Author: 45ck
  • Source: 45ck/skill-harness
  • License: MIT
  • Homepage: https://github.com/45ck/skill-harness/releases/tag/v0.1.0

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.