agentleFS
BagelHoleBagelHole/BagelSkills/skills/agent-proof-work/SKILL.md

agent-proof-work

Skill2 starsChanged 4 months agoMITView on GitHub

Use this skill when preparing a coding, research, operations, or content task for an autonomous AI agent. Convert vague intent into a bounded execution contract containing discoverable context, protected areas, acceptance criteria, verification commands, evidence requirements, escalation conditions, and a concise handoff format.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write task contracts autonomous agents can finish correctly without guessing hidden decisions or declaring victory early.

    line 20
  2. Distinctive:

    If a required input is retrievable from available tools or artifacts, retrieve it rather than asking the user.

    line 42
  3. Distinctive:

    Check that product decisions are settled enough to delegate.

    line 48
  4. Distinctive:

    Do not hide product decisions inside “use best judgment.”

    line 85

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Agent-Proof Work

Outcome

Write task contracts autonomous agents can finish correctly without guessing hidden decisions or declaring victory early.

This skill changes the agent's process: it requires evidence, explicit claim strength, and a checkable completion boundary. Load references/playbook.md only when deeper prompts or rubrics are needed. Use assets/output-template.md when producing the final artifact.

When to use

  • The user wants to delegate a task to Claude Code, Codex, Copilot, Gemini, Hermes, or another autonomous agent.
  • An issue, ticket, or prompt repeatedly produces incomplete or off-scope agent work.
  • A multi-step task needs reliable completion and handoff boundaries.

Do not use when

  • The user is directly asking the current agent to execute a simple task now.
  • The task is still strategically ambiguous; clarify or grill it before packaging.

Required inputs

  • Desired outcome and why it matters
  • Repository/system context and authoritative instructions
  • Scope, protected areas, constraints, and allowed side effects
  • Acceptance and verification expectations

If a required input is retrievable from available tools or artifacts, retrieve it rather than asking the user. Ask only when a missing answer represents authority, preference, or risk that tools cannot resolve.

Workflow

1. Resolve delegation readiness

Check that product decisions are settled enough to delegate. Separate unknowns the agent can discover from decisions only a human can make; escalate the latter before drafting.

Complete when: human-only decisions are resolved or explicitly escalated

2. Write the outcome contract

State observable end state, user impact, and acceptance criteria. Avoid prescribing an implementation unless the method is itself constrained.

Complete when: success can be observed without interpreting intent

3. Bound the search space

Name starting paths, authoritative documents, relevant systems, non-goals, protected areas, and forbidden side effects. Point to sources instead of pasting stale duplicates.

Complete when: the agent can discover efficiently without crossing boundaries

4. Define execution discipline

Require prerequisite inspection, incremental checks, repository conventions, dependency policy, and how to handle pre-existing failures. Specify whether commits, branches, or pull requests are expected.

Complete when: execution expectations match the host project

5. Define proof and escalation

Give executable verification commands where known, behavioral evidence requirements, and conditions for stopping to ask rather than guessing. Never include secrets in the task.

Complete when: success cannot be claimed without the requested evidence

6. Package the handoff

Produce a self-contained task prompt plus a compact completion report schema: outcome, changes, verification and exact results, artifacts, assumptions, and blockers.

Complete when: the prompt is self-contained and the report schema is concise

Operating rules

  • Specify outcomes precisely and implementations only when necessary.
  • Do not hide product decisions inside “use best judgment.”
  • Verification must test changed behavior, not only syntax or file existence.
  • Protected areas and side effects must be explicit for high-risk work.
  • The final task must stand alone without access to the drafting conversation.

Output contract

The final result must contain:

  • Self-contained agent task prompt
  • Outcome and acceptance criteria
  • Context entrypoints and protected boundaries
  • Verification commands and behavioral proof
  • Escalation/stop conditions
  • Completion report template

Distinguish observed, inferred, assumed, and unverified claims. Link to files, commits, screenshots, commands, logs, or supplied sources when those artifacts support a consequential conclusion. Keep the executive result concise; move detail behind appendices when necessary.

Common pitfalls

  • Prompt dumping: overwhelming the agent with context that has no bearing on decisions.
  • Over-specifying a brittle implementation while under-specifying the outcome.
  • Verification such as “tests pass” without commands or behavior coverage.
  • Assuming the delegated agent remembers prior conversation.

Verification checklist

  • The requested outcome—not merely the procedure—has been satisfied.
  • Every consequential claim is supported, qualified, or explicitly unverified.
  • The workflow's completion criteria are met in order.
  • Safety boundaries and forbidden side effects were respected.
  • The output follows assets/output-template.md or contains equivalent information.
  • Remaining unknowns have an owner, next action, or stated consequence.

Resources

  • references/playbook.md — deeper domain prompts and evaluation guidance.
  • assets/output-template.md — reusable final-output structure.
  • evals/evals.json — output-quality cases.
  • evals/trigger-evals.json — positive and negative description-trigger cases.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.