tdg-personal/autonomous-agent-harness
Transform Claude Code into a fully autonomous agent system with persistent memory, scheduled operations, computer use, and task queuing. Replaces standalone agent frameworks (Hermes, AutoGPT) by leveraging Claude Code's native crons, dispatch, MCP tools, and memory. Use when the user wants continuous autonomous operation, scheduled tasks, or a self-directing agent loop.
73
73%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong description that clearly articulates specific capabilities (persistent memory, scheduled operations, task queuing, computer use), names the technologies involved (crons, dispatch, MCP tools), and provides explicit trigger guidance via a 'Use when' clause. It also helpfully positions itself against known alternatives (Hermes, AutoGPT), which aids both distinctiveness and trigger matching. The description is concise yet comprehensive.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete capabilities: persistent memory, scheduled operations, computer use, task queuing, crons, dispatch, MCP tools. Also names concrete alternatives it replaces (Hermes, AutoGPT). | 3 / 3 |
Completeness | Clearly answers both what ('Transform Claude Code into a fully autonomous agent system with persistent memory, scheduled operations, computer use, and task queuing') and when ('Use when the user wants continuous autonomous operation, scheduled tasks, or a self-directing agent loop') with an explicit 'Use when' clause. | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms users would say: 'autonomous agent', 'scheduled tasks', 'agent loop', 'persistent memory', 'crons', 'AutoGPT', 'Hermes', 'self-directing', 'continuous autonomous operation'. Good coverage of how users would describe this need. | 3 / 3 |
Distinctiveness Conflict Risk | Occupies a very clear niche around autonomous agent systems, scheduled operations, and agent loops. The specific mention of crons, dispatch, MCP tools, and named frameworks (Hermes, AutoGPT) makes it highly distinctive and unlikely to conflict with other skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
35%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill attempts to cover an ambitious scope (autonomous agent system) but suffers from verbosity, speculative/unverifiable tool references, and a descriptive rather than instructive tone. The architecture diagram and Hermes comparison table consume significant tokens without adding actionable value. The example workflows read more like aspirational descriptions than executable instructions, and critical validation steps are missing throughout.
Suggestions
Cut the ASCII architecture diagram, Hermes comparison table, and capability descriptions — focus only on concrete setup steps and executable examples that Claude doesn't already know.
Replace pseudocode workflow descriptions (e.g., 'Autonomous PR Reviewer') with actual executable code or specific tool invocations that can be copy-pasted.
Add explicit validation checkpoints: after creating each MCP server config, verify it's running; after creating a scheduled task, confirm it was registered; include error recovery steps.
Split into SKILL.md (quick start + setup) with references to separate files for example workflows, MCP configuration details, and the cron patterns reference table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~200+ lines with significant padding. The ASCII architecture diagram, the Hermes comparison table, and explanations of concepts Claude already knows (what crons are, what memory is, what computer use does) waste substantial tokens. Many sections describe capabilities rather than instruct. | 1 / 3 |
Actionability | Some concrete code snippets exist (MCP config JSON, bash commands, cron patterns table), but much of the content is pseudocode or abstract workflow descriptions (e.g., 'Autonomous PR Reviewer' is numbered prose, not executable). The dispatch API endpoint appears fabricated, and MCP server package names may not exist, reducing real-world executability. | 2 / 3 |
Workflow Clarity | The Setup Guide provides a reasonable 4-step sequence, and example workflows list numbered steps. However, there are no validation checkpoints, no error recovery loops, and no verification that scheduled tasks were actually created or that MCP servers are running. The constraints section mentions error handling but doesn't show how to implement it. | 2 / 3 |
Progressive Disclosure | The content has clear section headers and some logical organization, but it's monolithic — everything is inline in one massive file. The architecture diagram, Hermes comparison table, example workflows, and setup guide could all be split into referenced files. No external references are provided for deeper dives. | 2 / 3 |
Total | 7 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 10 / 11 Passed | |
Reviewed
Table of Contents