Install
$ agentstack add skill-ariaxhan-the-agent-library-critique-vision-handoff ✓ 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
Critique Vision Handoff
Find the gaps that would make a coding agent waste time, guess wrong, or build the wrong thing.
Be direct and constructive. Every issue needs a consequence and a fix. Critique the what and why, not implementation choices the coding agent should own.
1. Establish the stakes
Identify the context:
- Hackathon/prototype
- Production system
- Library/API
- Internal tool
- Research experiment
If stakes are unclear, state the assumed stakes and calibrate severity around them.
2. Run the six critique layers
Structural integrity
Ask whether the document stands alone, stays internally consistent, includes required sections, and speaks to the intended audience.
Decision archaeology
Check whether decisions are specific, rationales are present, rejected alternatives are documented, and constraints have sources.
Complexity audit
Look for feature creep, excess components, unnecessary abstractions, and fragile dependency load.
Ambiguity detection
Flag weasel words, missing edge cases, implicit assumptions, and vague success criteria.
Risk surface
Check external failures, scale assumptions, security exposure, migration path, and operational failure modes.
Implementation readiness
Verify the first step, priorities, definition of done, and permission for the coding agent to push back.
3. Apply critique techniques
Use at least three:
- Naive Robot Test: read literally with zero context.
- Adversarial User Test: misuse or overload the system.
- Time Travel Test: debug it six months later.
- Subtraction Test: remove features and see what still works.
- Why Ladder: ask why three times for each major decision.
- Contradiction Hunt: compare sections for drift.
4. Write the critique
Use severity buckets:
BLOCKER: cannot proceed until fixedMAJOR: will cause implementation pain or wrong workMINOR: polish or clarity issueNITPICK: optional style issue
For each finding:
- Location
- Problem
- Why it matters
- Fix
Also include:
- What is good
- Simplification opportunities
- Questions the coding agent will ask
- Recommended next steps
5. Avoid bad critique
Do not:
- Nitpick formatting unless it harms understanding
- Fight explicitly documented technology decisions
- Add process that does not improve implementation
- Convert taste preferences into blockers
Before finalizing, ask: if the author fixes everything here, is the document actually better?
Use reference/techniques-and-output.md for the full output template and anti-pattern list.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: ariaxhan
- Source: ariaxhan/the-agent-library
- 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.