deepagents-harness — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited deepagents-harness (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Design long-running agents as a harness with explicit state, bounded delegation, durable artifacts, and reviewable tool calls. Use a higher-level harness when the work needs more than one ReAct loop.
| Need | Pattern |
|---|---|
| Short single-step assistant | Plain tool-calling agent |
| Stateful workflow with branches | LangGraph-style graph |
| Long-horizon work with files and subagents | Deep agent harness |
| Tool execution can affect users or infrastructure | Human-in-the-loop gate |
| Agent needs learned project practices | Durable memory plus context evolution |
Every subagent handoff should include:
DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, BLOCKED.See harness-contracts.md for reusable templates.
| Mistake | Fix |
|---|---|
| Letting subagents explore the whole repo | Give exact files, commands, and output contracts |
| Passing huge tool outputs through chat | Store them as artifacts and pass summaries |
| Treating checkpoints as logs only | Resume from checkpoints and test recovery |
| Adding memory without invalidation | Version memory and attach evidence |
| Evaluating only pass/fail | Inspect trajectory quality, retries, dead ends, and repair behavior |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.