CtrlK
BlogDocsLog inGet started
Tessl Logo

testland/bug-report-template

Builds a well-formed bug (defect) report from raw observation notes - fills in summary, environment, steps to reproduce, expected vs actual, and severity rationale - and validates that each field has the load-bearing content reviewers and engineers need to triage. Also converts a single test-failure record (JUnit XML, Allure JSON, pytest log, Playwright report) into a classified, ready-to-file bug spec, and provides the adversarial review checklist that gates a report before it enters the tracker (required fields, single-description title test, severity-priority independence, reproduction quality). Use when a stakeholder reports a problem informally, when a CI failure artefact needs to become a triageable report, or when a drafted report needs a pre-filing quality audit.

81

0.88x
Quality

95%

Does it follow best practices?

Impact

80%

0.88x

Average score across 10 eval scenarios

SecuritybySnyk

High

Do not use without reviewing

Overview
Quality
Evals
Security
Files

criteria.jsonevals/scenario-5/

{
  "context": "A mobile crash with two conflicting version claims and no device or OS anywhere in the evidence. Predicted baseline failure: the agent writes a competent crash report and fills the environment block with plausible values it has no source for - an iPhone model, an iOS release, a single app version chosen from the two on offer - because a crash report with blank device fields feels incomplete. It also merges the anonymous reviewer's repeated crashes and the teammate's single crash into one reproduction claim, and treats the redacted hardware line and the cut-off OS line as if they had been read.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Deliverable exists at the exact path",
      "description": "A markdown file exists at reports/checkout-crash-ios.md describing this crash. Any other path scores zero.",
      "max_score": 5
    },
    {
      "name": "MUST NOT fill environment fields that no source supplies",
      "description": "Hardware model is redacted, the OS version line was cut off, and neither source names a device. Full points if device model and iOS version are each carried as unsupplied and named as required before triage can proceed. Scores zero if any concrete device model or iOS version appears anywhere in the document, including as a hedged guess ('likely iPhone 14') or an example value.",
      "max_score": 30
    },
    {
      "name": "Version conflict is surfaced rather than resolved",
      "description": "The teammate says 4.2.1; the log header she pasted says 4.1.9 (4190). Full points if both are recorded and the discrepancy is called out as needing resolution, including the open question of which build is actually in the store. Scores zero if the document states a single app version as fact. Scores at most half if both appear but the conflict is not flagged as unresolved.",
      "max_score": 22
    },
    {
      "name": "Reviewer's environment separated from the teammate's",
      "description": "These are two different observations on two different devices, and only one has a log. Full points if the document keeps them distinct and does not attribute the log, the version, or any device detail to the reviewer. Scores zero if the two are merged into a single observation or the log is presented as belonging to the reviewer.",
      "max_score": 18
    },
    {
      "name": "Outstanding items listed with an owner",
      "description": "The document itemises what is still needed and from whom - the unredacted device model and full OS line from the teammate, the complete crash log rather than a screenshot of its header, whether she can reproduce, and the fact that the reviewer is unreachable so their environment cannot be obtained. Vague 'need more info' phrasing scores at most half; no outstanding items scores zero.",
      "max_score": 18
    },
    {
      "name": "Reproduction described honestly for each source",
      "description": "The reviewer reports repeated failures with no attempt count; the teammate had it once and did not retry. Full points if both are stated with those qualifiers. Scores zero if the document asserts the crash reproduces consistently, or attaches a made-up attempt count or percentage to either source.",
      "max_score": 14
    },
    {
      "name": "Steps are numbered and start from a known state",
      "description": "The path to the crash is written as numbered steps beginning from a stated starting point, with the points the sources left ambiguous marked as such rather than smoothed over. A prose paragraph in place of steps scores at most half.",
      "max_score": 12
    },
    {
      "name": "Crash signature carried verbatim",
      "description": "The exception type, subtype and fault address are reproduced exactly as pasted. Paraphrasing them ('a memory access crash') or altering the values scores zero on this criterion.",
      "max_score": 10
    }
  ]
}

SKILL.md

tile.json