Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Watch For Harness Mode

ASecurity

Use when a proven watch pattern should become a reusable harness artifact with configurable inputs. Not for remote, credential, publish, deploy, or irreversible changes.

35 stars
0 votes
0 copies
0 views
Added 9/12/2026
ai-agents

Works with

terminalcli

Security Analysis

A100/100

Scanned 9/12/2026

Install to Claude Code

$npx -y skills add OutlineDriven/odin-claude-plugin --skill watch-for-harness-mode --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Watch For Harness Mode?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Watch For Harness Mode
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/outlinedriven-watch-for-harness-mode-odin-claude-plugin/badge)](https://www.skillsdirectory.com/skills/outlinedriven-watch-for-harness-mode-odin-claude-plugin)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: watch-for-harness-mode
description: 'Use when a proven watch pattern should become a reusable harness artifact with configurable inputs. Not for remote, credential, publish, deploy, or irreversible changes.'
disable-model-invocation: true
---

# Watch for harness mode

## Contract

| Field | Bound contract |
|---|---|
| Trigger | A watch pattern has proven itself and the user wants to turn it into a reusable harness. |
| Authority | Reversible local: writes only the named harness artifact file; rollback is deleting or reverting the harness artifact. No VCS commit, remote call, credential use, or published artifact. |
| Side effect | One harness artifact as a reversible local state change. |
| Done | A reusable harness is created from the proven watch. |

## Inputs

- Watch transcript (required): the proven watch pattern: a session transcript, run log, artifact, or notes that document the watch's target, interval, evaluation logic, completion predicate, and observed success.
- Harness name (required): a slug for the harness artifact file.
- Harness description (optional): a one-sentence description of what the harness checks and when it succeeds. Defaults to "Reusable harness derived from a proven watch pattern."
- Default poll interval (optional): the polling interval in seconds. Defaults to the interval observed in the watch transcript, or 60 if unspecified.
- Default deadline (optional): the maximum polling duration in seconds. Defaults to the deadline observed in the watch transcript, or 300 if unspecified.
- Adoption confirmation (required): the user explicitly confirms or declines to adopt the harness artifact.

## Procedure

1. Validate inputs. Reject if the watch transcript is absent or contains no observable target, completion predicate, or evaluation logic. Stop if the harness name is absent or contains path-separator characters. Done when: inputs are validated, or the step has stopped on ambiguous inputs.
2. Bound scope. The harness artifact is the only write target. No VCS commit, no remote call, no credential use, no published artifact. If the harness name already exists as a file in the working directory, stop and report the collision. Done when: scope is bounded and no name collision exists.
3. Extract the watch mechanism from the transcript: (a) the target surface being watched, (b) the completion predicate that determined success, (c) the evaluation logic that applied the predicate, (d) the poll interval, and (e) the deadline. Done when: all five elements are extracted, or the step has stopped reporting the missing element.
4. Synthesize the harness. Write a harness artifact named `{harness-name}.md` (or the user's preferred extension) containing: the harness name, the extracted description or default, the extracted target, the extracted completion predicate, the extracted evaluation logic, configurable input fields for the target, completion predicate, poll interval, and deadline with their extracted defaults, and the adoption confirmation step. Done when: harness artifact is written with all extracted elements and configurable fields.
5. Present for adoption. Show the harness artifact to the user. Ask for explicit confirmation or declination. Do not assume adoption. Done when: user response is received.
6. Confirm or withdraw. If the user confirms: mark the harness as adopted and report success. If the user declines: delete the harness artifact and report withdrawal. If the user requests modifications: record the requested changes and stop. This skill produces an initial draft, not an iterative editor. Done when: adoption is confirmed, withdrawal is completed, or modification request is recorded and the step has stopped.

## Failure and recovery
- Missing watch mechanism: the watch transcript does not contain a target, completion predicate, or evaluation logic. Partial extraction is not sufficient. Stop without writing an artifact.
- Harness name collision: a file with the harness name already exists in the working directory. Stop without overwriting. Report the collision and the conflicting path.
- User declines: the user explicitly declines to adopt the harness artifact. Delete the artifact. Report withdrawal as the terminal result. Do not retry, re-prompt, or represent this as a failure.
- User requests modifications: the user asks for changes to the harness artifact after seeing it. Stop without modifying. The initial draft is the contract of this skill.
- Non-converged watch: the watch transcript documents a failed or non-converged watch. Stop. Report that the watch pattern was not proven and a harness cannot be derived from it. Do not produce a harness from a non-converged watch.

## Output
One terminal result: adopted (harness artifact exists and user confirmed, with name and path), withdrawn (user declined, artifact deleted, no harness exists), or blocked (input validation failed or watch transcript insufficient, with the blocking reason).

Attribution

OutlineDrivenOutlineDriven
View sourceMore from OutlineDriven →
SSkills DirectorySkills Directory

Know which skills are safe — weekly.

Best new skills + every skill we flagged as malicious. From the team that scanned 103,619.

Join free

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Know which skills are safe — weekly.

Best new skills + every skill we flagged as malicious. From the team that scanned 103,619.

Join free

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

651 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →