markusdowne/handoff-integrity-check
Validate agent handoff packets and resume readiness using schema, freshness, and replay checks. Use when tasks pause/resume across sessions, agents, or humans — including when a user wants to continue where they left off, hand off to another agent, resume a previous task, or pick up an interrupted workflow.
91
90%
Does it follow best practices?
Impact
95%
1.25xAverage score across 4 eval scenarios
Passed
No known issues
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 an excellent skill description that clearly articulates specific capabilities (handoff packet validation with schema, freshness, and replay checks), provides comprehensive trigger guidance with natural user language, and occupies a distinct niche. The description is concise yet thorough, uses proper third-person voice, and includes both the 'what' and 'when' components effectively.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: 'validate agent handoff packets', 'resume readiness', 'schema checks', 'freshness checks', 'replay checks', and 'prompt-injection guardrails'. These are concrete, well-defined operations. | 3 / 3 |
Completeness | Clearly answers both 'what' (validate handoff packets using schema, freshness, and replay checks with prompt-injection guardrails) and 'when' (explicit 'Use when...' clause covering multiple trigger scenarios like pausing/resuming, handing off, continuing interrupted workflows). | 3 / 3 |
Trigger Term Quality | Excellent coverage of natural trigger terms users would say: 'pause/resume across sessions', 'continue where they left off', 'hand off to another agent', 'resume a previous task', 'pick up an interrupted workflow'. These are phrases users would naturally use. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche — agent handoff validation with schema/freshness/replay checks is a very specific domain unlikely to overlap with other skills. The combination of handoff packets, cross-session resumption, and prompt-injection guardrails creates a clear, unique identity. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-crafted skill with strong actionability and workflow clarity — the sequenced validation steps, concrete examples (both valid and invalid), executable Python code, and clear classification criteria make it highly usable. The main weaknesses are minor verbosity (some repeated explanations of resume_token semantics) and the fact that all content is inline rather than leveraging progressive disclosure for the lengthy examples. Overall, it's a solid, production-ready skill.
Suggestions
Consolidate the resume_token guidance — the 'Interpret resume_token' section and the workflow's resume_token validation bullet overlap; merge them into one place to save tokens.
Consider moving the two full output examples into a separate EXAMPLES.md file and referencing it, keeping just one brief example inline to demonstrate the format.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is mostly efficient but has some redundancy — the invalid example with its explanation could be tightened, and some bullet points restate things Claude would infer (e.g., explaining what a plain continuity marker is multiple times). The untrusted-content guardrails section, while valuable, includes some obvious guidance for Claude. | 2 / 3 |
Actionability | Provides a concrete JSON schema, valid and invalid examples, executable Python code for freshness checking, specific classification criteria, and copy-paste-ready output formats. The workflow steps are specific and directly executable. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced with explicit validation checkpoints: schema validation → freshness check → resume_token validation → replay check → resume-boundary note → classification. The classification tiers (clean/operational/critical) provide clear decision points, and the resume-boundary note acts as a verification step before proceeding. | 3 / 3 |
Progressive Disclosure | The content is well-structured with clear sections and headers, but it's somewhat long for a single file (~120 lines of substantive content). The two full output examples and the valid/invalid packet examples could potentially be split into a reference file. However, there are no references to external files, which is acceptable given the skill's scope but means everything is inline. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
Reviewed
Table of Contents