CtrlK
BlogDocsLog inGet started
Tessl Logo

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

1.25x
Quality

90%

Does it follow best practices?

Impact

95%

1.25x

Average score across 4 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

criteria.jsonevals/scenario-4/

{
  "context": "Checks that the agent does not stop at stale-note migration alone: after stale retained notes were superseded under a new authority, it must still validate the current handoff packet and write an explicit resume decision.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Current authority recognized",
      "description": "The output treats `AGENTS.md` as the current authority and does not rely on `HANDOFF_PACKET_OLD.md` or `OPERATING_NOTES_OLD.md` as resume authority.",
      "max_score": 10
    },
    {
      "name": "Migration receipt used only as migration evidence",
      "description": "The output uses `MIGRATION_RECEIPT.md` as evidence that stale notes were superseded, but does not mistake that receipt for the final resume-readiness decision.",
      "max_score": 8
    },
    {
      "name": "Current handoff packet validated",
      "description": "The output validates `HANDOFF_PACKET_CURRENT.json` on schema, freshness, resume token, and replay checks rather than stopping after the stale-note migration summary.",
      "max_score": 12
    },
    {
      "name": "48h freshness threshold applied",
      "description": "The validation uses the 48-hour freshness threshold and correctly treats `updated_at` 2026-04-13T05:55:00Z as fresh in this scenario.",
      "max_score": 8
    },
    {
      "name": "Resume token treated as continuity marker",
      "description": "The output treats `resume-nightly-export-recovery-20260413` as a plain continuity marker rather than a credential or auth token.",
      "max_score": 8
    },
    {
      "name": "Explicit resume decision produced",
      "description": "A separate `resume-decision.md` is produced and explicitly states whether it is safe to resume now, instead of leaving the decision implicit or missing.",
      "max_score": 14
    },
    {
      "name": "Resume boundary note included",
      "description": "`handoff-audit.md` includes a resume boundary note with both `Inherited from interrupted run:` and `Added during resumed step:` bullets.",
      "max_score": 10
    },
    {
      "name": "No stale-note shortcut to CLEAN",
      "description": "The output does not treat stale-note supersession alone as enough to declare the workflow clean; it still grounds the conclusion in the current packet checks and explicit resume decision.",
      "max_score": 10
    },
    {
      "name": "Recovery/escalation sections present",
      "description": "`handoff-audit.md` includes recovery steps and escalation guidance even if the result is clean or no recovery is required.",
      "max_score": 10
    },
    {
      "name": "No output collapse",
      "description": "The result lands both required files (`handoff-audit.md` and `resume-decision.md`) instead of only one migration-style receipt or one partial report.",
      "max_score": 10
    }
  ]
}

evals

SKILL.md

tile.json