← Back to Skills Marketplace
as76

Tenacity — Never Give Up

by Andrea Sassi · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
44
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install tenacity
Description
Keeps working autonomously on complex or long tasks without stopping for confirmation, only pausing when genuinely blocked or at completion.
README (SKILL.md)

Tenacity — Autonomous Goal Completion

This skill implements the Tenacity execution pattern: the agent keeps running until the stated goal is reached, only asking for human input when genuinely blocked.

Core Principle

Default behavior: Ask for confirmation only when blocked, not at every step. If the path is clear, execute.

When to Activate

Activate when Andrea says things like:

  • "keep going until done"
  • "don't stop to ask"
  • "run this autonomously"
  • "finish it completely"
  • Any long-running multi-step task

Architecture

Tenacity uses OpenClaw's existing primitives:

Component Role
Standing Orders Grants permanent authority to execute within scope
Task Flow Durable multi-step state across gateway restarts
Lobster Deterministic steps with resume tokens
Cron Scheduling and enforcement
Checkpoint State saved after each milestone

Execution Pattern

Step 1: Define Scope (Standing Order)

Before starting, establish:

  • What the agent is authorized to do
  • When to stop and ask (hard blockers only)
  • What constitutes completion
## Standing Order: [Task Name]

**Authority:** [exactly what the agent can do]
**Hard blockers:** [only conditions that require human input]
**Completion:** [how to know when done]
**Escalation:** [when to alert if something goes wrong]

Step 2: Create Task Flow

Use a persistent session (session:\x3Cid>) for the task flow so state survives restarts:

openclaw tasks create "Tenacity: [task]" --session session:tenacity-task

Or use cron with a named session:

openclaw cron add \
  --name "[Task] tenacity run" \
  --session session:tenacity-task \
  --message "Execute [task] per standing orders. Check checkpoint before resuming." \
  --announce \
  --channel telegram \
  --to "834732674"

Step 3: Checkpoint Protocol

After each milestone, save state:

CHECKPOINT: milestone_X_completed
STATE: {step: 3, last_file: "output.json", errors: []}
TIMESTAMP: 2026-05-14T10:30:00Z

On restart, resume from checkpoint:

RESUME from checkpoint: milestone_X_completed
Last state: {step: 3, last_file: "output.json"}
Continue from step 4...

Step 4: Hard Blocker Conditions

Ask only when:

  1. File/system permission denied and no alternative path
  2. External dependency unavailable after all retries
  3. Decision required that changes scope or direction
  4. Andrea explicitly set a constraint

Never ask when:

  • A step can be attempted with alternative tools
  • A command failed but a retry or workaround exists
  • The path is clear but requires multiple steps
  • Output could be improved but is already acceptable

Step 5: Completion

When goal is reached:

  • Save final checkpoint with STATUS: COMPLETE
  • Brief summary to Andrea
  • Log to memory/tenacity-log.md

Checkpoint Script

See scripts/checkpoint.sh — run after each milestone:

bash scripts/checkpoint.sh "step_3_done" '{"step": 3, "last": "file.csv"}'

Resume Protocol

On session start, check for incomplete checkpoints:

bash scripts/checkpoint.sh --resume

If resume point exists, announce: "Resuming from [milestone]..."

References

  • Task Flow: docs/automation/taskflow.md
  • Standing Orders: docs/automation/standing-orders.md
  • Cron: docs/automation/cron-jobs.md
  • Lobster: See openclaw lobster --help
Usage Guidance
Install or invoke this only if you want a highly autonomous agent for a narrowly defined task. Before using it, remove the hardcoded Telegram recipient, define exact allowed files/tools/actions, require approval for deletion and command execution, set a clear stop condition, and clean up cron jobs, checkpoints, and memory logs afterward.
Capability Analysis
Type: OpenClaw Skill Name: tenacity Version: 1.0.0 The 'tenacity' skill is designed to bypass human-in-the-loop (HITL) confirmations, granting the agent broad autonomous authority to execute commands, modify files, and send messages without intervention. This significantly increases the impact of potential prompt injection attacks. The 'scripts/checkpoint.sh' script is vulnerable to path traversal via the milestone ID, and 'SKILL.md' includes a hardcoded Telegram ID (834732674) for notifications, which could serve as a pre-configured exfiltration or control channel.
Capability Assessment
Purpose & Capability
The autonomous-completion purpose is coherent, but the artifacts broaden it to any long-running task and permanent standing orders, which is high-impact without enough built-in containment.
Instruction Scope
The skill changes the agent's normal stopping behavior by telling it to avoid confirmation except for hard blockers and to continue through failures when workarounds exist.
Install Mechanism
There is no install spec or external dependency, and the included checkpoint script only reads/writes checkpoint JSON under /tmp by default; the main risks are behavioral instructions rather than hidden install code.
Credentials
The standing-orders template allows reading, writing, editing, deleting workspace files, running commands, web searching, messaging, and using any workspace tool without asking.
Persistence & Privilege
The skill explicitly uses persistent sessions, standing orders, cron, checkpoints, and memory logs so tasks can resume and continue across restarts, with no clear expiry or cleanup requirement.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install tenacity
  3. After installation, invoke the skill by name or use /tenacity
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Prima release: esecuzione autonoma senza conferme intermedie. Attivato da 'keep going', 'finish it', 'run until complete', 'autonomous tenacity'.
Metadata
Slug tenacity
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Tenacity — Never Give Up?

Keeps working autonomously on complex or long tasks without stopping for confirmation, only pausing when genuinely blocked or at completion. It is an AI Agent Skill for Claude Code / OpenClaw, with 44 downloads so far.

How do I install Tenacity — Never Give Up?

Run "/install tenacity" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Tenacity — Never Give Up free?

Yes, Tenacity — Never Give Up is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Tenacity — Never Give Up support?

Tenacity — Never Give Up is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Tenacity — Never Give Up?

It is built and maintained by Andrea Sassi (@as76); the current version is v1.0.0.

💬 Comments