Harness Skill Proposal
Propose new or revised harness skills and policy text from recurring failure patterns while keeping scope limited to harness assets.
DocumentationRequirements & limits
Use when repeated harness failures justify drafting new or revised harness skills and policy text.
Inputs
- reflection findings
- current harness policies
- current skills affected by the failure pattern
Outputs
- Proposed skill changes
- Boundary notes
- Rationale summary
Before you begin
- Reading and copying the procedure require no terminal setup. Check requirements for any packaged programs separately.
Try this first
Read the procedure and supply its documented inputs. Use the full text in an agent app or chat that supports pasted instructions, following that host's guidance.
Limits for this step
- Instruction package only; it does not provide or authorize an executable runtime capability.
- Safe for proposal drafting inside harness-owned assets; require human approval before any change outside harness scope.
Read full instructionsHide full instructions
Harness Skill Proposal
When to use
- Use after run reflection identifies repeated harness failures.
- Use when a new harness rule or skill is needed.
- Use before creating a harness-only PR.
Inputs required
- reflection findings
- current harness policies
- current skills affected by the failure pattern
Workflow
- Read the reflection evidence and isolate a recurring pattern.
- Decide whether the fix belongs in a skill, a policy file, or a benchmark.
- Draft the proposed change in a structured, reviewable format.
- Confirm the target stays within harness-owned paths.
Output format
- Problem Statement
- Proposed Harness Change
- Target Artifacts
- Expected Improvement
- Approval Notes
Guardrails
- Only target harness-owned artifacts by default.
- Do not propose edits to production code or deployment config.
- Tie every proposal to concrete run evidence.
- Escalate if the best fix would require protected-path changes.
Failure modes
- If evidence is too thin, recommend more benchmarks instead of a change.
- If multiple fixes are plausible, present the smallest safe proposal first.
- If scope expands beyond harness paths, stop and require approval.
View file metadata
---
name: harness-skill-proposal
description: Turn recurring harness failures into candidate skill or policy
updates while keeping proposals limited to harness-owned artifacts.
---Install instruction files
These commands copy guidance into the selected runtime directories. Confirm that your agent app discovers it. Any packaged programs need separate setup and verification.
Install (Codex)
./bin/skills-hub install agentops/[email protected] --runtime codex
Install (Claude)
./bin/skills-hub install agentops/[email protected] --runtime claude
Install (Generic)
./bin/skills-hub install agentops/[email protected] --runtime generic --target ./my-agent/skills