Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Hitl Approver — measured AI agent skill | DecimalAI

stijnman/hitl-approver

SkillScore88Maintained

Pauses high-risk actions for explicit human approval. Use before WhatsApp se nds, deployments, deletions, credential changes, or financial actions. Trigg ers:approve this, hitl check, human review needed. Use when the user needs t his capability.

by stijnman

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer pack

decimalai skills pull stijnman-hitl-approver

Get this skill — curl · raw file

curl -fsSL https://app.decimal.ai/s/stijnman-hitl-approver/SKILL.md

Raw file: /s/stijnman-hitl-approver/SKILL.md ↗ · machine-readable summary at /s/stijnman-hitl-approver.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

24 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 7d ago

Yes — +63% lift-34% tokens0% turns88% of whole cases pass with skill (21/24)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	-59%	0%
case-02	✗→✓	▲ Improved	-51%	0%
case-03	✗→✓	▲ Improved	-39%	0%
case-04	✗→✓	▲ Improved	-37%	0%
case-05	✗→✓	▲ Improved	-41%	0%

Showing 5 of 24 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

stijnman / grok-custom-skills

.grok/skills/hitl-approver/SKILL.md

View on GitHub ↗

License MITImported 8/1/2026Version v1

Skill contents

SKILL.md

Hitl Approver

When to Use

User says approve this or task matches this capability
User says hitl check or task matches this capability
User says human review needed or task matches this capability
User says confirm before send or task matches this capability

Workflow

Classify action: message / deploy / delete / financial / credential.
Present briefing: summary, risk (low/medium/high), rollback plan.
Wait for explicit approval (yes / approve / go ahead).
On approval, execute and confirm outcome. On denial, abort and suggest safer path.

Output Template

HITL Required: {type}
Summary: {summary}
Risk: {level}
Rollback: {plan}
Approve? (yes/no)

Integrations

whatsapp-auto-responder
privacy-redactor
multi-platform-messenger-bridge

Error Handling

| Failure | Response | |---------|----------| | Ambiguous approval | Re-ask; never infer consent from silence. | | Timeout in session | Defer action; remind user once. | | User says no | Abort; log reason; offer alternative. |

Gotchas

Medium+ risk always requires HITL even if user seems hurried.

Example

Input: User request matching triggers above. Output: Structured result per workflow with integrations invoked as needed.

Want to see if stijnman/hitl-approver would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.