Index of every page on this site: llms.txt
All skills
Run it yourself

Audit or design agent-facing product capabilities: action/context parity, discoverable contracts, safe mutations, verifiable outcomes, and recovery. Use for explicit agent-readiness reviews of apps, repos, CLIs, APIs, MCP/A2A integrations, or Agent Skills; not every general code or documentation task.

Use this Skill: https://skilld.dev/gh/tristanmanchester/agent-skills/agent-use

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesfiles-mobile-and-long-running-work.md

≈394 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Files, mobile, and long-running work

Files as universal interface

Files are often the simplest shared workspace. Good file-based agent surfaces have stable paths, clear ownership, explicit generated-file markers, sidecar metadata where needed, atomic writes, diffs/previews, and validation commands.

Good patterns:

  • AGENTS.md documents where agents may edit.
  • Generated files are clearly marked and reproducible.
  • Long outputs are written as artifacts and linked.
  • State files use JSON/YAML/CSV/Markdown with schemas where possible.
  • Checkpoints are durable and recoverable.

Self-modification

Self-modifying systems need guardrails: version control, branches, tests, rollback, human review for risky changes, changelogs, and clear distinction between policy/prompt changes and executable code changes.

Mobile and offline constraints

Mobile agent surfaces must account for battery, network loss, OS background limits, permissions, storage sandboxes, and user-visible approval. Work should checkpoint before suspension, resume safely, avoid unbounded background activity, and expose status when the app foregrounds.

Long-running jobs

Long tasks need a lifecycle contract: create/start, status, logs/events, artifacts, checkpoint, cancel, resume, completion signal, error envelope, and final verification.

Minimum task object:

{
  "task_id": "task_123",
  "status": "running",
  "progress": 0.4,
  "checkpoint": "after-validation",
  "resume_token": "res_abc",
  "artifacts": [],
  "next_action": "poll status or wait for webhook"
}

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning1mo3 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    The skill 'agent-use' is a comprehensive toolkit for auditing and designing agent-ready systems. It includes documentation, templates, and Python scripts for scanning repositories and websites. Security analysis found no malicious intent; identified network and command execution patterns are aligned with the skill's primary purpose as an auditing tool.

  • Socket1mo

    1 alert: gptAnomaly

  • Snyk1mo

    Risk: LOW · No issues

Signed by skilld at 3323bc9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Steadyupdated last month
metadata
{
  "version": "3.0.0",
  "reviewed": "2026-09-13"
}
Other metadata
compatibility
Local helper scripts require Python 3.10+; structural validation also requires PyYAML 6.0.3 or later in major 6. Network and execution checks need separately authorised tools and environments.

README badge

README badge for tristanmanchester/agent-skills/agent-use